How do the best marathon runners in the world train? While you might catch a workout or two on Instagram or hear rumors about epic training weeks on message boards, there’s precious little information on the systematic approaches that elite coaches use with top marathon runners–and even less information on the science that backs up these approaches for designing marathon training programs.
One exception to this general rule has been the Italian coach Renato Canova, arguably the greatest living running coach and the topic of several of my previous posts on Running Writings. Canova freely discusses his training philosophy and posts example workouts or even full training schedules for the athletes he has worked with, which include Olympic and World Championships medallists.
In 1999, Canova even co-authored a book on the science of marathon training—however, there’s a bit of a catch: this book was printed through the IAAF (now World Athletics), not a traditional publishing company or printing press. As a result, Canova’s book is extraordinarily rare. I had heard of this book probably a decade ago, but in the intervening years I couldn't find any substantive information on its contents. Until now.
I was recently able to acquire a copy of Marathon Training - A Scientific Approach by Enrico Arcelli and Renato Canova. At least according to WorldCat, the copy I read is the only institutionally-held copy of this book in North America (incidentally, it’s at NC State of all places; no word on whether Laurie Henes has read it).
What follows is my in-depth summary and review of Renato Canova’s book. In the summary, I’ll present the information as reported in the book.
I’ll refrain from injecting my own opinion or additional information until the second part of this article, where I’ll compare this book (which is now almost 25 years old) with some of Canova’s more recent comments on marathon training, plus add some additional scientific and coaching context on adapting these workouts to mid-level and sub-elite runners for those interested.
Technical details or tangential points are relegated to hyperlinked footnotes at the bottom of the article.[1]
This is a long article! You can use the link icons 🔗 next to each heading to bookmark your progress, or to point a friend to a specific section of this article when you send them the link.
Hopefully this approach will allow readers who disagree with my own interpretations to see for themselves the case that Arcelli and Canova make in this book for their system, while also allowing for some consideration of what’s changed in the intervening years. Canova himself has said in an interview that if he were to rewrite this book (even back in 2011!), he would change 50% of what he wrote.
Even though a lot has changed since 1999, this book is still an absolute treasure trove–the training methods it details worked magnificently for Gelindo Bordin (gold medal at the 1988 Olympic marathon), Ornella Ferrara (1995 World Championships marathon bronze medal), and Stefano Baldini (who at the time of the book’s writing had run 2:07 in the marathon; he subsequently won the 2004 Olympic marathon), to name just a few.
Though Renato Canova is now more closely associated with the success of Kenyan and Ethiopian marathon runners, these results are solid evidence the same techniques can work for all high-level marathon runners.

Authors and introduction: Coaches Renato Canova and Enrico Arcelli 🔗
In addition to Renato Canova, Marathon Training - A Scientific Approach was co-authored by Enrico (Henry) Arcelli, an Italian physician who authored many books and articles on training, and served as the head coach for the Italian Track and Field Federation. Sadly, Arcelli passed away in 2015. Arcelli’s physiology knowledge complements Canova’s coaching knowledge, and puts a physiological basis underneath their marathon training strategy.
Canova and Arcelli emphasize that training is not a simple formula or set of rules, but a set of principles that relate how specific types of training affect specific physiological mechanisms inside the body.
The right combination of training creates the right set of physiological adaptations in the body to support marathon performance (notably, these adaptations might be different for different runners!). A good understanding of physiology is framed as a way to “achieve a rational training method for marathon runners,” not as the means for an ironclad prescription of a specific set of workouts.
Chapter 1: Physiological aspects of the marathon runner’s performance 🔗
Chapter 1 begins with an overview of the energetics of ATP production in the muscles, which supplies the energy required for muscles to produce force (I can’t help but point out that this is also how I chose to open my own book on training!). By far the most important energy pathways discussed are the glycolytic system and the aerobic system.
Energy systems in marathon running 🔗
Glycolysis, sometimes called the “anaerobic system,” is the first step in the breakdown of carbohydrates into energy. Glucose, a simple sugar, is broken down into pyruvate, and later lactate, a process which yields a small amount of energy, and a small amount of acidity (via H+ ions). Importantly, this system can operate without oxygen, but the H+ ions accumulate and cause fatigue. The glycolytic system is used when muscles need to produce force but do not have sufficient energy to do so using the aerobic system.
Canova and Arcelli argue that the importance of the glycolytic system (which they term the “anaerobic lactic system”) is underestimated by most marathon coaches–though the marathon is too long to rely heavily on anaerobic energy, the ability to “shuttle” and reprocess lactate in other muscles is an important capability to build in marathon runners (a topic discussed subsequently in later chapters).
The aerobic system can break down either carbohydrates or fats. The aerobic system does not cause fatigue via the accumulation of H+ ions, but it does require oxygen to function. When the energetic demands of exercise exceed the aerobic system’s capacity, muscles are forced to rely on glycolysis, leading to the familiar, rapid buildup in acidity and fatigue when running at high speeds.
The aerobic system’s ability to oxidize either carbohydrates or fats is an important consideration for marathon training, as specific types of workouts can target each of these respective sub-components of the aerobic system.
Because the aerobic system requires oxygen to function, the components of the body that deliver oxygen to the muscles (and the mitochondria in the muscles that use that oxygen) are critical determinants of marathon performance. Arcelli and Canova break down these factors into central and peripheral components of the aerobic system.
The central components of the aerobic system include the heart, lungs, and blood. Canova and Arcelli argue that the limiting central factor for trained distance runners tends to be stroke volume, pointing to evidence that maximum heart rate varies widely among athletes of similar abilities and that most athletes are not limited by diffusion of oxygen from the lungs into the blood.
When it comes to the peripheral components of the aerobic system, the two most important components are the capillary networks in the muscles, which deliver oxygen to muscle fibers, and the levels of specific enzymes found in the mitochondria inside muscle fibers, which put oxygen to use in regenerating ATP.
As will be familiar to any student of physiology, the combined effects of the central and peripheral components of the aerobic system determine a runner’s maximal oxygen uptake, or VO2 max.
Lactate dynamics in marathoning 🔗
A large part of Chapter 1 covers the generation, transport, and oxidation of lactate. Even at “fully aerobic” speeds, lactate is generated inside muscle fibers, then shuttled out into the blood and subsequently oxidized (aerobically!), either by adjacent muscle fibers, other muscles, the heart, liver, or kidneys.
The greater the speed, the greater the amount of lactate produced–again, even at speeds we nominally think of as fully aerobic and even when blood lactate concentrations are constant. Beyond a certain speed, though, the body’s ability to transport and oxidize lactate is overwhelmed by the continued production of greater and greater amounts of lactate at faster speeds, and the concentration of lactate in the blood starts to climb.
Canova and Arcelli cover the typical range of lactate concentrations in the blood, from 1 mmol/L[2] at rest or during slow running, to 2 mmol/L at marathon pace or at the aerobic threshold, to the familiar 4 mmol/L level associated with the lactate threshold (or race pace for a one-hour race), to 18-25 mmol/L, levels only found in high-level athletes after middle-distance races.
Canova and Arcelli then describe the anaerobic threshold:[3] it is the fastest pace at which lactate levels in the blood exist in equilibrium. A runner’s pace at the anaerobic threshold is more strongly associated with marathon performance than maximal oxygen uptake, but increasing VO2 max tends to increase the anaerobic threshold as well. The aerobic threshold gets a brief mention, but is not formally defined (other than corresponding to a blood lactate level of 2 mmol/L and being very close to marathon pace).
Canova and Arcelli then briefly detail the different types of muscle fibers: type I (slow-twitch) fibers, type IIa (fast-twitch oxidative), and type IIb (fast-twitch glycolytic).
The fast-twitch Type IIa and IIb fibers are differentiated by their ability to use the aerobic system to generate energy: fast-twitch oxidative fibers have a greater capacity to regenerate ATP using the aerobic system, and are also more amenable to responding to aerobically-based training, compared to Type IIb fibers. Canova and Arcelli also note that many Type II fibers are of an intermediate type, and can become more slow-twitch-like with training.
Effects of increasing speed on lactate and muscle fiber recruitment 🔗
The effects of increasing speeds on the various components of the runner’s body are illustrated in a table showing how the body responds to a range of paces, from a very slow jog at about 50% slower than marathon pace up to a very fast run at 30% faster than marathon pace (circa 1500m pace).[4] At slow speeds, blood lactate levels are low, the bulk of the caloric cost of running is generated by oxidizing fat instead of carbs, and only slow twitch muscle fibers are recruited.
As running speed increases, muscle fiber recruitment shifts, with more fast twitch oxidative fibers being recruited at moderate speeds, to the fast twitch glycolytic fibers being recruited at the fastest speeds. Carbohydrate utilization increases as well, reaching 100% at 15% faster than marathon pace (circa 5k race pace for a 2:25 marathon runner). Blood lactate levels climb accordingly, first staying at an equilibrium up until about 105% of marathon pace, then climbing steadily from thereon out.
Unit cost and running economy in marathoners 🔗
Canova and Arcelli then discuss the “unit cost” of running, more commonly called the “metabolic cost of transport” or “running economy.” Unit cost is defined as the amount of oxygen required to cover one kilometer of running. It’s not mentioned explicitly in the book, but it’s a well-established physiological principle that unit cost is essentially constant as a function of speed, i.e. a 6-minute mile has the same energetic cost (per unit distance) as a 10-minute mile.
Citing scientific data from a 1985 comparative study of marathoners who were grouped into "elite" (~2:21 marathon), "good" (~2:37), and "slow" (~3:24) categories, Canova and Arcelli point out that unit cost tends to be lower in the faster marathoners, though there is significant overlap in unit cost among the groups: some elites had a higher unit cost than the slow runners, and vice versa.
Two points regarding unit cost warrant further emphasis.
First, though a wide range of marathon performances are possible with a given unit cost, the energetic demands (and therefore the carbohydrate consumption) for a runner with a high unit cost are likewise higher, meaning less economical runners run a greater risk of running out of fuel late in the marathon.
Second, some–but not all–runners show a significant increase in their unit cost late in a marathon, possibly because of glycogen depletion.
Fat oxidation and marathon training 🔗
Section 1.6 is quite brief, but makes a point I had not seen in previous writing on marathon training–drawing on the physiology covered already, Canova and Arcelli point out that because the energetic cost of running per unit time increases as a function of speed, but the relative contribution of fat to energy output decreases as a function of speed, there exists a pace at which the absolute rate of fat oxidation–which the authors call “aerobic fat power”–is maximized.
This pace occurs at 85-90% of anaerobic threshold pace, or 90-95% of marathon pace. Doing long, fast runs at this pace is an effective way to increase this capability. Run too fast, and the relative contribution of lipids (fats) shrinks to zero; run too slow, and the absolute energy demand of running is too low. This is the strongest physiological justification for the “long fast run” as a core element in marathon training that I’ve encountered in the training literature.
Lactate as an additional energy reservoir in the marathon 🔗
We now return to the understated importance of lactate as a fuel in the marathon. Canova and Arcelli call lactate the "fifth reservoir" of energy, with the first four being glycogen in the muscle, glucose from the liver, triglycerides in the muscles, and fatty acids from fat cells.
Citing the “lactate shuttle” research by George A. Brooks, the authors emphasize that an important factor in marathon performance is training the slow twitch fibers to oxidize lactate produced in adjacent fast-twitch fibers. This effect is particularly important later in the race, when the slow twitch fibers have exhausted their glycogen stores and would otherwise be unable to continue producing force.
The specific adaptation necessary in the slow twitch fibers is an increase in the H-LDH (lactate dehydrogenase) enzyme, which facilitates the conversion of lactate into pyruvate, for subsequent use by the aerobic energy system.
The most important determinants of marathon performance 🔗
After a brief summary of the basics of heat regulation and hydration, Chapter 1 concludes with an analysis of how the various systems discussed above contribute to marathon performance.
Maximum oxygen uptake (VO2 max) is highlighted as an important determinant of marathon performance, with Canova and Arcelli writing that increases in VO2 max tend to increase the speed that marathoners can sustain for several hours in a race. However, they also point to the example of Derek Clayton, a 2:09 marathoner whose VO2 max was <70 mL/kg/min, as evidence that factors beyond maximal oxygen intake influence marathon performance.
Unit cost, or running efficiency, is one of these additional factors. At fast speeds, highly economical runners use nearly 20% less oxygen per kilometer than less-economical ones. Canova and Arcelli mention body fat as a factor as well, since both running economy and maximum oxygen uptake are expressed relative to body mass–overweight runners should lose body fat, if it can be done without affecting the qualities of their muscles.
Beyond VO2 max and unit cost, the anaerobic threshold is another major factor in marathon performance, so much so that knowing an athlete’s anaerobic threshold allows a reasonably good prediction of marathon pace. Canova’s guideline for identifying the anaerobic threshold is that it should be about 5% faster than marathon pace, regardless of an athlete’s ability level.
Finally, Canova and Arcelli highlight that carbohydrates alone are not sufficient to fuel a marathon, and as such, aerobic fat power is an additional determinant of marathon success.
Chapter 2: Physiological aspects of the marathon runner’s performance 🔗
Chapter 2 is concerned with designing workout types to target specific physiological adaptations.
This chapter covers broad trends in workouts–the specifics about the actual workouts (how fast, how long, etc.) are covered in Chapter 4.
Renato Canova and Enrico Arcelli detail strategies for targeting the various physiological systems that influence marathon performance. Interestingly, even though these goals share much in common with American-style physiologically-based training programs (e.g. Daniels’ Running Formula), the workouts chosen to elicit the desired physiological adaptations are quite different.
I find this fascinating, since it’s sometimes taken as “common wisdom” that Daniels-style workouts are the best or only way to improve VO2 max and lactate threshold. Canova and Arcelli clearly feel differently. Canova and Arcelli appear to be early adopters of the training implications of George A. Brooks’ lactate shuttle hypothesis, which as far as I can tell did not enter into the American coaching sphere until much later.
Chapter 2 begins with a reminder that our goal in training is to induce long-term adaptations in the body that are specific to marathon performance–adaptations beneficial for the 400m or the shot put would be detrimental for the marathon. Instigating these marathon-specific biological adaptations requires the right type and duration of training stimulus.
A stimulus needs to “embarrass” the right biological system; as an example, Canova and Arcelli point to aerobic fat power, which will not be stimulated if the intensity of a run is too high, because of the declining contribution of fat to energetic consumption discussed above.
Increasing VO2 max in marathon runners 🔗
The first example of divergence from American-style physiologically-based training comes in the section on improving the central components of the aerobic system, specifically stroke output. Canova and Arcelli write that the most effective training they have found for increasing stroke output (and therefore VO2 max) is fast uphill sprints lasting 15 seconds or less, on very steep hills (15% gradient or steeper!).[5]
Their rationale for these hill sprints comes originally from research by the pioneering German physiologist Herbert Reindell, who connected adaptations in stroke output with the rate of increase in heart rate at the beginning of a repetition in an interval workout.
Notably, this goal of a rapid heart rate increase is quite different from the American style, which aims to sustain maximum cardiac output for as long as possible–both within a rep, and overall during a workout. The American approach gives rise to higher-volume workouts at 3k to 5k pace, such as 8x800m at 3k pace with 2-3 min jog recovery.
A typical session might include “a few” sets of ten sprints, with each at "near maximum intensity." No hard rules on recovery between reps are given, except that heart rate should return to 120-130 beats per minute between reps. Because these hill repeats trigger a very rapid rise in heart rate, Canova and Arcelli claim, this type of session induces strong adaptations in stroke output.[6]
Increasing oxygen uptake in muscle fibers 🔗
Moving on to the peripheral components of the aerobic system, the next topic discussed is how to increase the ability of the mitochondria to put oxygen to use during high-intensity running.
Pointing to research by Italian mitochondria researcher Gianni Benzi, Canova and Arcelli write that the best way to generate an increase in the levels of aerobic enzymes in mitochondria is, paradoxically, to slightly exceed the aerobic capabilities of the body, pushing into speeds that elicit small amounts of lactate accumulation.
This principle is in alignment with the general theme of "embarrassing" the body to generate a new adaptation to a stimulus, and is another important point on which Canova-style training differs from traditional American training.
Canova and Arcelli recommend paces from 97% to 105% of the anaerobic threshold, either as a fast continuous run, or as interval workouts with reps of up to a few kilometers in length. Different paces within this 97-105% range elicit slightly different responses in muscle fibers of different types–on the slow end of the range, the primary stimulus is limited to the slow twitch fibers, whereas paces fractionally above the anaerobic threshold improve the aerobic capabilities of the fast twitch oxidative fibers as well.
Increasing the duration of a threshold workout is another way to target a greater proportion of the fast twitch oxidative fibers: high-volume sessions will eventually exhaust the slow-twitch fibers, forcing fast-twitch oxidative fibers to rotate in to continue the effort.
Steady uphill runs can also generate a fast-twitch-oriented response, though the appropriate intensity is more difficult to get right. Canova and Arcelli write that a heart rate monitor can be used to ensure the effort remains within 97-105% of anaerobic threshold.[7]
Increasing lactate uptake and oxidation in muscles 🔗
The next section covers how to develop the ability to transport and oxidize lactate via an increase in the levels of lactate dehydrogenase (H-LDH). Canova and Arcelli recommend two types of sessions to train this ability:
- Fast continuous runs at speeds between the aerobic threshold and the anaerobic threshold.
- Repeats of up to a few kilometers in length at or slightly faster than anaerobic threshold pace, with recovery runs that are initially at a slow speed, but as training progresses, eventually build to fairly fast speeds (approaching the aerobic threshold).
According to Canova and Arcelli, it is this type of training that pushes the aerobic threshold very close to the anaerobic threshold–if the amount of lactate entering the blood stays the same, but the amount of lactate being taken up and oxidized increases, a marathon runner will be able to sustain the same speed with a lower blood lactate level.
Boosting lipid consumption rate in marathon runners 🔗
Chapter 2 closes by returning to the idea of aerobic lipidic (or fat) power, discussed earlier. Predictably, the best way to increase the aerobic fat power is to run for a long time at the speed that maximizes fat oxidation.
The speeds that maximize aerobic fat power are between 92 and 100% of marathon pace. Importantly, exceeding marathon pace (even if you are still below the anaerobic threshold) will fail to produce the right stimulus for lipidic power, because the proportion of your energy coming from fat is too low.
Conversely, running too slow is not desirable either, since your absolute energy expenditure is too low. Less-experienced runners can start at paces slower than 92% of marathon pace, but as they progress the speed will have to be increased to continue improving their aerobic fat power.
Chapter 3: Evaluation tests 🔗
Chapter 3 is relatively brief, and is concerned with various field tests that can be used to determine marathon pace and various other training paces, like anaerobic threshold pace.
One implied takeaway from this chapter is that workout paces (e.g. percent of marathon pace or % MP) should be based on the current fitness of the runner, not their goal time, though this is never explicitly stated in the book.
Predicting marathon performance from half-marathon times 🔗
The easiest test proposed by Canova and Arcelli for determining marathon pace is to simply add 5% to a runner's half-marathon time (so marathon pace is 95% of HM pace). For runners unable to sustain 95% of HM pace for the full marathon, Canova and Arcelli attribute insufficient aerobic fat power capabilities as the cause–the discussion on slow-twitch versus fast-twitch marathoners in Chapter 5 is also relevant here).
Predicting anaerobic threshold (lactate threshold) with the Conconi test 🔗
The Conconi test[8] is a field test that uses a heart rate monitor and a carefully-controlled progression of speeds[9] to determine the heart rate (and speed) that corresponds to the anaerobic threshold, a.k.a. the lactate threshold.
The test works like this: first, do a warm-up. Then, on a track, you start running at a slow pace (9:40-8:00/mi, though remember Canova works primarily with elite athletes!), increasing the speed by 0.5 km/hour every 200 meters. That corresponds to a 2 sec/200m increase in pace at 7:00/mi, and a 1 sec/200m increase at 5:00/mi. Canova and Arcelli seem to suggest that the exact splits are not particularly important, as long as the pace is steadily progressing each 200m.
The outcome of interest in the Conconi test is the "deflection velocity", which is determined by making a plot of heart rate as a function of running speed using the 200m splits and the heart rate taken by the watch, then determining when the heart rate/speed relationship deviates from linearity. The deflection velocity is equal to the anaerobic threshold speed (and heart rate).
As with many official-sounding physiology tests, actually identifying this deflection velocity more or less tends to come down to “eyeballing it,” as shown on the graph below.[10]

Plot of heart rate on each 200m split as a function of running speed. Red line shows a linear fit to the data below 17 km/hr, and the blue line shows a smooth fit to all of the data.[11]
Determining aerobic and anaerobic (lactate) threshold from blood lactate meter data 🔗
Canova and Arcelli also provide a protocol for using a lactate meter to determine the anaerobic and aerobic thresholds, which is becoming increasingly accessible to everyday runners due to decreases in the cost of lactate analyzers and the increasing popularity of blood lactate testing, in part due to its use by the Ingebrigtsen family and other adherents of the Norwegian school of lactate training.[12]
The protocol involves a series of 2000m repeats (at least four, often more), starting slightly slower than an athlete’s estimated marathon pace, and working down to somewhat faster than anaerobic threshold pace. Canova and Arcelli emphasize that these repeats must be extremely evenly paced (you can use a tempo trainer to accomplish this). Wearing a heart rate monitor during this test is a good idea, since you can use it to establish your heart rate at anaerobic threshold pace.
Each 2k repeat should only increase in pace by 2.5 to 5 seconds per kilometer. Unfortunately, the book fails to mention how much rest you should take between reps. My experience with lab testing is that it should be short–probably 30-60 sec is going to be necessary to take an earlobe lactate measurement, but you don’t want to wait more than a few minutes, otherwise your blood lactate will drop and you’ll get a falsely-low reading (though the long duration of the reps is a pretty solid safeguard against this issue).

The plot above from data in the book depicts results for one athlete (Athlete A, in blue) completing 5x2000m, starting at 3:30/km and dropping 10 sec/km each rep.[13] This athlete’s lactate threshold appears to be around 3:10-3:05/km. Since Canova and Arcelli write that lactate concentrations of between 2 and 3 mmol/L are sustainable for the marathon, this runner’s marathon fitness is likely between between 3:20 and 3:15/km.
The second dataset illustrates another runner (Athlete B, in red) whose 7x2000m session indicates around 3:01/km as the fastest pace sustainable for the marathon, and a lactate threshold of around 2:48/km.
The chapter concludes with some brief discussion of oxygen consumption testing for determining aerobic fat power, but concludes that it isn’t very reliable due to measurement errors in wearable oxygen consumption monitors.[14]
Chapter 4: Training means for the marathon runner 🔗
Chapter 4 finally gets to workout specifics: what types of sessions, how long, how fast, and when in training? Unlike Chapter 2, this chapter is organized by workout type, not by physiological target—in part because one type of workout can have multiple physiological targets.
When specific examples are given of workouts and their effects, I’ll describe them in each section, with details and notes on implementation when available.
See also the table of Canova's workouts for a useful overview of how to Canova and Arcelli prescribe workouts across the full range of speeds during training.
Weekly mileage for marathon runners 🔗
Canova and Arcelli write that their top men run 230-250 km (143-155 mi) per week, with peaks of up to 280-300 km (174-186 mi) per week (!), and top women should cover "no less than" 190 to 230 km (118-143 mi) per week (!!). They write that a "large portion" of this training should be at about 90% of marathon pace.
No details are provided in the text on exactly what portion of mileage should be at this steady pace, or how long these runs should be, though they do note that they can be beneficial even when they are relatively short.
Steady paced continuous runs 🔗
These workouts are pretty simple: X duration at Y pace. Canova and Arcelli write that the range of paces and durations mediate which physiological systems are targeted by a continuous run.
Paces close to anaerobic threshold influence oxygen uptake in the muscles and lactate reuptake ability. In contrast, running at 90-100% MP influences aerobic fat power and lactate oxidation. When running at 80% of marathon pace, direct effects on performance are limited, and become even more so for slower paces.
Fast continuous run
20-40 minutes at 97-100% of anaerobic threshold, or equivalently, 104-107% of marathon pace.
Effects: Increasing oxygen uptake in the muscles, and improving lactate oxidation in the muscles. Can also be substituted with a cross country race or a road race over 8-15 km.
Continuous run at marathon pace
18-25 km at 100-103% of marathon pace during general training, or 30-35 km at 97-100% of marathon pace during the specific phase
Effects: These workouts are useful for developing a sense of the pace that is sustainable for the marathon. The longer version is a marathon-specific workout and should be done in the final 6-8 weeks before the marathon.
Steady medium-fast runs
45-80 min at 95-97% MP during general training, or 32-36 km runs during the specific phase
These runs are only described in the table, and are not mentioned in the main text, but they are an obvious missing link between runs at 100-103% MP and 90-92% MP.
Medium runs and marathon endurance runs
Canova and Arcelli recommends 2-3 hour run durations at 85-90% MP, and 60-90 minute runs (or the occasional 40-52 km run during the specific phase) at 90-92% MP.
Effects: The primary benefit of these types of runs is to increase aerobic fat power, and induce adaptations in both fat cells and muscle fibers. These training sessions are also effective when done as doubles (because intramuscular lipid stores aren't fully restored in between training sessions).
Progressive runs 🔗
Canova and Arcelli distinguish a range of types of progressive runs, ranging in duration from 20 to 90 minutes. The following examples are given:
Fast progressive runs
20-40 minutes, starting at 102% marathon pace and progressing to 108% of marathon pace.
Often done "stepwise", e.g. 12 km broken into 4x3k at 100 / 102 / 104 / 106% of MP, as a workout for a 2:08 marathoner.
Medium fast progressive runs
45 to 60 minutes, progressing from about 95 to 105% of marathon pace.
An example workout given is 20 min at 96% MP, 20min at 100% MP, and 15 min at 105% MP for a 2:28 (female) marathoner.
Medium progressive run
60 to 90 minutes, progressing from 85 to 100% of marathon pace.
As an example, Canova and Arcelli give 90 minutes with the first 30 min at 87% MP, the second 30 min at 91% MP, and the final 30 min at 96% MP, again for a 2:08 marathoner.
Effects: Canova and Arcelli write that all types of progressive runs are especially effective at inducing oxygen uptake in fast twitch fibers. The progressive nature of the run accelerates the rate at which muscle fibers are "rotated out" of use, which amplifies the stimulus to the fast twitch fibers.
Renato Canova's special blocks for marathon runners 🔗
The "special block" is one of Canova's most famous innovations. It's actually discussed in the previous section on progressive runs, since marathon special blocks have progressive aspects to them.
The special block is a morning/afternoon workout pair, both of which involve significant portions of fast running. Each session involves 10-15 km at 85-90% of marathon pace, followed by one of three different workout types. These three categories are "intensive-extensive", "extensive-intensive", and "extensive."
Intensive-extensive special blocks
This type of special block workout emphasizes speeds at 105-110% marathon pace for the second part of both the morning and evening blocks. Canova and Arcelli give an example from Maria Curatolo's training prior to her silver medal at the European Championships:[15]
AM: 10 km at 90% MP + 10k at 106% MP
PM: 10 km at 87% MP + 10x1000m at 111% MP with 2min recovery
Extensive-intensive special block
This type of special block workout emphasizes continuous running at marathon pace. Canova and Arcelli give the example of Davide Milesi's training prior to his third-place finish in the 1995 World Cup in Athens:[16]
AM: 10 km at 89% MP + 15 km at 101% MP
PM: 10 km at 90% MP + 15 km at 101% MP
Extensive special block
This type of special block workout emphasizes longer continuous running at fractionally slower than marathon pace (around 98% MP). Canova and Arcelli give an example from Ornella Ferrara's training just before her bronze medal at the 1995 World Championships.[17]
AM: 15 min easy + 24 km at 99% MP
PM: 15 min easy + 24 km at 100% MP (see percentage note above)
Runs with pace variations for marathon runners 🔗
Another type of workout that Renato Canova is quite famous for is alternating-pace workouts. In their book, Canova and Arcelli outline a range of possible workouts of this type, noting that they often become nearly synonymous with more traditional interval workouts in some cases:
- Long pace alternations totaling 15-21 km with 5-7 km intervals at 103-107% MP. Example: 3x5k at 102%, 103%, and 104% with 2-3min “slower” recovery in between.
- Medium pace alternations totaling 12-15 km, with 3-5 km intervals at 1-5-108% MP. Example: 5k-4k-3k at 104%, 106%, 108% MP with 3 min “slower” recovery in between.
- Short pace alterations totaling 10-12 km with 1-3k intervals at 106-110% MP. Example: 10x1k at 109% MP with 2min “slower” recovery in between
- Mixed pace alterations totaling 10-12 km, with intervals from 400m to 3000m at 107-112% MP. Example: 3k at 109% MP, 3min recovery; 2k at 111% MP,[18] 3min recovery; 10x400m at 115% MP, 1min recovery
- Long run with short pickups totaling 105-135 min of running with the steady portion at 80% MP, with 500-1000 m pickups at 103-105% MP. Example: one hour steady + 10x1.5 min faster, 1.5min slower, then 30 min steady at the end.
- Long run with long pickups totaling 105-135 min of running with the steady portion at 80% MP with 3-7 km pickups at 100% MP. Example: 30min steady + 7k-5k-3k at 100% MP, with 10min slower in between, then 20-40 min steady at the end
Repetition runs (interval workouts) for marathon runners 🔗
Getting back to more traditional workouts, the following section provides copious details on different types of interval workouts (called “repetition runs” in the book). Canova and Arcelli report that the three most useful types of interval sessions for marathoners are:
- Repetitions of 300-500m at up to 10% faster than the anaerobic threshold
- 1-3 km repeats at close to anaerobic threshold pace
- Repeats of 3-7 km at slightly faster than marathon pace, with the recoveries only slightly backed off from marathon pace.
These are described in detail in the following sections.
Fast intervals
Canova and Arcelli write that these sessions are effective at boosting maximum oxygen uptake, via the same rapid increase in heart rate employed in the fast uphill sprints.
Example: 15-20 (or more) x 300-500m at 110% MP to 110% anaerobic threshold (which would be ~115% MP for an athlete good at both the marathon and the 10k).
Anaerobic threshold intervals
Canova and Arcelli recommend 1-3 km repeats at 97-105% of anaerobic threshold pace, noting that these sessions stimulate the production of aerobic enzymes in the mitochondria.
The appropriate pace is faster for shorter repeats (e.g. 100-105% of threshold for 1 km reps, but only 97-100% of threshold for 3 km reps), and the total workout volume (including recovery) should be around 20 km. Importantly, the recovery should be running, not walking or standing, to facilitate circulation of blood and reuptake of lactate by other muscle fibers. No specific workout examples are given.
Marathon pace intervals
These are the core race-specific sessions done in the final period of training before the marathon. The most notable feature of these workouts is that the recovery is quite fast–only a bit slower than marathon pace. Canova and Arcelli point out that marathon pace intervals can be considered an extension of anaerobic threshold intervals, dialing the distance up and backing off the pace a bit.
As with special blocks, marathon pace intervals can be of the “extensive” type or the “intensive” type, as shown with the following examples:
Example (extensive): 3x7k at 103-105% MP with 1 km recovery at 98% MP
(From Gelindo Bordin’s training before his Seoul Olympic marathon victory)
Example (extensive): 4x6k at 102-104% MP with 1 km recovery at 96-97% MP
(From Ornella Ferrara’s training before her 1995 World Championships bronze medal)
Example (intensive): 10x1k at 105% MP with 1 km recovery at 100-101% MP
(Three weeks prior to Stefano Baldini’s 2:07:57 at the 1997 London Marathon)
Example (intensive): 20x500m at 107% MP with 500m recovery at 95-96% MP
(19 days prior to Maria Curatolo’s silver medal at the 1994 European Championships)
Hill workouts and continuous uphill runs for marathon training 🔗
Like progressive runs, hill workouts are particularly effective for stimulating improvements in aerobic capabilities of the fast-twitch oxidative muscle fibers. Canova and Arcelli lay out the following types of hill sessions:
Short hill sprints: Short hill sprints of ~60m in length on very steep hills (~15% grade) should be performed relatively far out from the marathon, and are more useful for slow-twitch dominant runners whose 5k and 10k performances are not as strong as their marathon times (see the discussion above on hill sprints for improving oxygen uptake)
Medium hill reps: Little detail is provided on these, aside from that they should be 300-1000m long on inclines of 5-10% grade, and that their benefit is in increasing aerobic enzymes in the muscles.
Continuous uphill runs: Lasting 6-10 km on slopes of 3-6%, these runs stimulate fast twitch oxidative fibers and as such are useful for more fast-twitch type runners who have the tendency to slow down significantly later in the marathon. Earlier in the book, the authors suggest that a heart rate equal to 97-105% of anaerobic threshold is the correct intensity for this workout.
Runs in hilly terrain: As with medium hill reps, little detail is provided on these except that they increase muscle strength and vary the recruitment of muscle fibers because of the changing slope. Elsewhere in the book, these are described as “long” runs through undulating hills, with no further qualifiers on “long.”
How to use the different marathon workouts 🔗
The chapter closes with some useful summary tables of all of the various workouts, grouped by their primary training benefit. Canova and Arcelli aggregate these benefits into three main goals: boosting oxygen uptake in the muscle fibers (which they call “aerobic power”), boosting lipid consumption (which they call “aerobic endurance”), and increasing marathon-specific fitness.[19]
I’ve compiled these workouts into the charts below and combined them with information elsewhere in the book for easier reference. Below, the term "general phases" refers to both the introductory and fundamental phases jointly.
Workout paces and distances for Renato Canova’s marathon training[20] 🔗
Pace (% MP) | Workout type |
>110% MP | Short intervals, e.g. 10x500m with 90 sec recovery, then 10x400m with 60 sec recovery |
108-110% MP | 1-3 km repeats totaling 10-12 km |
105-108% MP | General phases: 20-40 min progressive Specific phase: 2-5k repeat totaling 12-16 km |
103-105% MP | General phases: 20-40 min continuous Specific phase: 3-7k repeats totaling 15-21 km |
97-100% MP | General phases: 45-80 minutes progressive or 18-25 km continuous Specific phase: 28-30 km continuous |
95-97% MP | General phases: 45-80 minutes continuous Specific phase: 32 - 36 km continuous |
92-95% MP | General phases: 60-90 min progressive Specific phase: 36-45 km continuous |
90-92% MP | General phases: 60-90 min continuous Specific phase: 40-52 km continuous |
85-90% MP | 120-180 min continuous |
80-85% MP | "Slow runs" |
<80% MP | Very easy “regeneration” runs |
Renato Canova’s workouts for increasing oxygen uptake in marathoners[21] 🔗
Workout | Distance and pace | Example |
Fast continuous run | 20-40 min at 104-107% MP | 10k at 103-104% MP |
Fast progressive run | 20-40 min at 102→108% MP | 12k with 3k segments at 100/102/104/106% MP |
Long pace alternations[22] | Reps of 5-7 km at 103-107% MP, totaling 15-21 km | 3x5k at 102/103/104% MP with 3 min recovery |
Medium pace alternations | Reps of 3-5 km at 105-108% MP, totaling 12-15 km | 5-4-3k at 104/106/108% MP with 3 min recovery |
Short pace alternations | Reps of 1-3 km at 106-110% MP, totaling 10-12 km | 10x1k at 109% MP with 2 min recovery or 5x2k at 107% MP with 3 min recovery |
Mixed pace alternations | Reps of 400-3000m at 107-112% MP, totaling 12-15 km | 3k at 109% MP, 4min recovery, 2k at 111% MP, 3min recovery, 10x400m at 115% MP, 1 min recovery |
Continuous uphill run | 6-10 km uphill on a 3-6% grade | 8 km fast uphill at ~100-103% anaerobic threshold heart rate[23] |
Race | 6-12 km competition on roads, cross country, or track | 10k race at 107% MP or 5k race at 110% MP |
Renato Canova’s workouts for increasing fat oxidation in marathoners[24] 🔗
Workout | Distance and pace | Example |
Medium fast progressive run | 45-60 min at 95→105% MP | 20-20-15 min at 96/100/105% MP |
Medium progressive run | 60-90 min at 85→100% MP | 30-30-30 min at 87/91/96% MP |
Steady medium pace run | 60-90 min at 92-90% MP | N/A[25] |
Steady long run | 120-180 min at 85-90% MP | N/A |
long runs with short pickups | 105-135 min at 80% MP, with 500-1000m pickups at 103-105% MP | 60 min at 80% MP + 10x1.5 min at 103-105% MP w/ 1.5min recovery + 30mi at 80% MP |
Long runs with long pickups | 105-135 min at 80% MP with 3-7km pickups at 100% MP | 30 min at 80% MP + 7-5-3k at 100% MP with 10min slower recovery + 20-40 min at 80% MP |
Long run through hilly terrain | N/A[26] | N/A |
Renato Canova’s marathon specific workouts[27] 🔗
Workout | Distance and pace | Example |
Continuous run at marathon pace | 18-25 km at 100% MP | Half marathon race at 100% MP |
Extensive marathon pace intervals | 19-30 km with 2-7 km reps at 100-102% MP with 1 km recovery at 85-95% MP | 4x5k at 101% MP with 1k recovery at 93-96% MP |
Intensive marathon pace intervals | 15-23 km with 500-1000m reps at 103% MP with 500-1000m recovery at 97% MP | 8x1k at 104% MP with 1k recovery at 98% MP or 20x500m at 103% MP with 500m recovery at 97%[28] |
Specific long runs | 30-35 km at 96-98% MP[29] | 32 km at 98% MP or 35 km at 96% MP |
Extensive-intensive special block | 10 km at 85% MP + 10-15 km at 100-103% MP, in morning and afternoon | AM: 10 km at 88% MP + 12 km at 101% MP PM: 10 km at 88% MP + 12 km at 101% MP |
Chapter 5: Training schedules for the individual marathon runner 🔗
Chapter five is divided into two parts. The first part is concerned with individualizing training for fast-twitch type versus slow-twitch type runners. The second part is about how training is periodized throughout a training cycle.
Training for fast-twitch vs. slow-twitch marathon runners 🔗
Canova and Arcelli describe two categories of marathoners: “enduring” marathoners, whose muscles are dominated by slow-twitch Type I fibers, and “fast” marathon runners, whose muscles have a greater proportion of Type IIa (fast-twitch oxidative) fibers.
Note that these are relative proportions; according to the authors, even in fast twitch marathoners, slow twitch fibers are still two-thirds of the total fiber composition (though for enduring runners, it can be as high as 90%). True fast-twitch runners with a high proportion of Type IIb fibers are not discussed, probably because they are not capable of high-level marathon performance.
The authors do note that many of the Type II muscle fibers are of an intermediate type, and can change their properties to become more fast-twitch-like or slow-twitch-like through training.
Identifying fast-twitch and slow-twitch marathoners
Fast-twitch marathon runners tend to succeed in 5k and 10k races, perform well in traditional interval workouts, and have an anaerobic threshold that tends to be significantly faster (>5%) than their marathon pace. These runners tend to run with a more “elastic” stride, were generally successful in the 3000m and even the 1500m in their high school years, but struggle at long, fast runs. They also recover more slowly following a marathon competition.
Slow-twitch marathon runners, in contrast, tend to struggle at 5k and 10k races and in faster workouts. In their high school years, they likely fared better in longer races as opposed to the middle distances. They have less than a 5% difference between their anaerobic threshold and their marathon pace, and tend to run with a flatter stride.
Training differences for fast-twitch versus slow-twitch marathoners
The differences between fast and slow-twitch marathoners induce some slightly counter-intuitive strategies for training.
For the fast-twitch marathoner, the primary goals for better marathon performance are not increasing VO2 max or lactate threshold–they are the ability of the fast oxidative muscle fibers to use oxygen, and the ability of their slow-twitch muscle fibers to reuptake and oxidize lactate generated by adjacent fast-twitch muscle fibers.
It’s a mistake for fast-twitch runners to do too much lactic work at speeds significantly above the anaerobic threshold, because this enhances what Canova and Arcelli call the “lactic features” of the muscles–levels of enzymes involved in glycolysis, which both accelerates fatigue accumulation and burns through glycogen stores. Instead, they should put more emphasis on doing workouts at paces ranging from 85-100% of anaerobic threshold (with particular focus on 95-100% of threshold), and increase the volume of these workouts over time.
For the slow-twitch marathon runner, VO2 max and lactate threshold are major barriers to running faster, and more emphasis should be shifted to targeting speeds that are at and above anaerobic threshold. For these athletes, acquiring “lactic features” in their muscles is much less of a concern, and they should incorporate higher intensities, with a focus on increasing oxygen uptake, on a more frequent basis, so that their training is more well-balanced in its distribution of intensities (compared to a fast-twitch runner).
Marathon training and periodization under Renato Canova 🔗
In the final part of the book, Canova and Arcelli outline their periodization strategy for a marathon training cycle. This strategy incorporates three phases: introductory, general, and specific training.[30]
Canova’s introductory phase for marathon training 🔗
The introductory phase is focused on re-acquiring the “capacity to work” and developing the general aspects of fitness that were (intentionally) neglected during the final months of preparation for the previous marathon.
The introductory phase lasts 6-8 weeks, and is initiated following a transitional period of rest and recovery after a previous race. Canova and Arcelli note that sometimes the turnaround between marathon races is too short, and the introductory phase must be shortened or even eliminated entirely.
The two goals during the introductory phase are the development of muscle efficiency and aerobic endurance.
Muscle efficiency is never explicitly defined, but appears to be a catch-all for muscle strength, flexibility, and running form. Canova and Arcelli recommend strength exercises, circuit training, proprioceptive training, technical drills, uphill sprints, flexibility exercises, and stretching to improve this quality.[31]
Developing aerobic endurance happens with very straightforward continuous runs. Canova and Arcelli highlight three types of workouts:
- Slow paced runs,[32] building over time to 90 minutes.
- Medium paced runs,[33] progressing over time to become both longer and faster. Ultimately these should progress to 45 minutes in duration by the end of the introductory phase.
- Continuous progressive runs, starting slow at the beginning and progressive to a “medium” pace. As with the other workouts, these increase in duration and speed over the course of the introductory period, ultimately reaching 60 minutes in duration.
Jointly, the introductory phase and the fundamental phase that follows are considered “general training,” and stand in contrast to the specific phase that comes after. Like the introductory phase, the fundamental phase is focused on building up the athlete’s capabilities, not preparing specifically for the race. It is, in some sense, “training to train,” as opposed to the “training for the race” which happens in the specific phase.
Canova’s fundamental phase for marathon training 🔗
Following the introductory phase, the fundamental phase is initiated and lasts 8-10 weeks. During this period, mileage reaches its highest level, and a variety of workouts are introduced to “attack” the body with various stimuli. Runners reach a state of general fatigue familiar to anyone who’s done high mileage before, which is a normal and healthy response to training.
Canova and Arcelli outline four goals: increasing aerobic power, developing anaerobic endurance, continuing to increase aerobic endurance, and maintaining the muscle efficiency developed in the introductory phase.
The definition of aerobic power in this chapter conflicts with how it was defined in earlier chapters. In Chapter 4, “aerobic power” corresponded to maximal oxygen uptake, and corresponded to training sessions at 102-112% MP. Here, though, it seems to refer to the concept of aerobic efficiency introduced in the introductory period, with workouts at paces in the vicinity of 90-92% MP.[34]
Indeed, the recommended workouts are extensions of the 45 minute medium pace run in the introductory phase. This session should bifurcate into two different types of workouts; one lasting a similar to slightly longer duration (45-75 minutes) while increasing in speed, and a second focused on increasing the duration at the same speed, extending the duration up to 60-120 minutes.
Developing anaerobic endurance has two components: increasing the ability to run at the anaerobic threshold, and increasing the ability to run faster than anaerobic threshold.
Running at the anaerobic threshold is developed via many of the workouts introduced in Chapter 4, specifically fast continuous runs, fast progressive runs, and continuous uphill runs. See the Chapter 4 summary for more details on these workouts.
Running above anaerobic threshold is given less discussion, but is developed via short uphill sprints and fast intervals. Again, see the Chapter 4 summary for more.
Muscle efficiency is, again, not discussed in detail, but probably involves similar techniques as the introductory phase.
Canova’s specific phase for marathon training 🔗
The specific phase of marathon training lasts the final 6-8 weeks before the race. Canova and Arcelli write that the highly demanding workouts in the specific phase, and the marathon itself, cause a deterioration in the “mechanical engine” and “nervous reservoir” of the athlete. This deterioration is why runners need a transitional, introductory, and fundamental phase of training following a marathon competition.
During the specific phase, Canova and Arcelli recommend competing rarely, if at all, and having only modest performance goals for these races. In their words it is “nearly impossible” to achieve high-level performances in shorter races too soon before the marathon.
During this phase, training becomes more “modulated,” with greater day-to-day variation in training volume, to ensure that the athlete recovers well after the major workouts, and arrives well-rested (physically and mentally) for the next marathon specific workout.
Like the description of the fundamental phase, understanding the specific phase is similarly muddied by inconsistent terminology; the book says that the main physiological targets are “power endurance,” “aerobic endurance,” and “aerobic power,” though these are not described in detail.
The book refers the reader to the table of marathon specific workouts detailed in Chapter 4; I will follow suit by directing you to a similar table of Canova's marathon-specific workouts in this article. Unfortunately, there is no guidance on how to structure or progress these marathon-specific workouts over time, and there are no concrete examples of what a training schedule for the specific phase would look like.
Discussion, commentary, and takeaways 🔗
Marathon Training - A Scientific Approach is a fascinating window into the coaching and scientific rationale behind Renato Canova’s approach to marathon training. Moreover, it’s a snapshot of the training systems that were first proven in Italy, before Canova started working primarily with Kenyan and Ethiopian runners–the “Kenyans are different” argument I sometimes hear when discussing Canova’s contemporary training idea doesn’t hold water when it comes to the workouts in this book.
Make no mistake, this book is definitely not perfect–the recommended field testing protocols for lactate threshold are somewhat dated, information on the different types of workouts is spread out over multiple chapters, and there are some significant misprints in the tables. In addition, the terminology for workouts and physiological targets is inconsistent, which adds some confusion. The misprints and inconsistent use of terms make it challenging to follow how to assemble the different workouts into a coherent training plan.
The biggest omission, though, is an example of a full day-by-day training schedule for an entire season that shows how workouts are distributed, and what kind of volume and recovery surrounds the long, marathon-specific workouts sessions.
Now, I shouldn’t complain too much about the lack of a full schedule, since Renato Canova has made the full training schedule of Gelindo Bordin before his 1988 Olympic marathon victory freely available! I have it bookmarked and will analyze it in light of the training approach laid out in this book later this year; I’ll post a link here when that’s finished.
Because of its various quirks, this book will be most useful for readers who are already somewhat familiar with Canova’s overall training approach. I wouldn’t recommend it as a first read in marathon training. Without some context on how to organize workouts week to week, it might feel like an overwhelming variety of workouts, paces, and types of sessions, without much guidance on how to incorporate them into an overall training program.
Even so, this is still a valuable book if you can find it. Writing a book is hard work, and I wish more coaches would do it–I could name a dozen coaches off the top of my head that I would love to see write a book on their training approach, even if that book turns out a bit disorganized or missing some tables and figures (heck, my own book had two “Chapter 9”s for its first two years of publication!). Canova and Arcelli deserve enormous praise for putting their training approach to paper in a systematic way.
My biggest takeaways 🔗
Marathon Training - A Scientific Approach made me think much more carefully about several important facets of marathon training. My three biggest takeaways from this book are:
- Training to improve fat oxidation rate–or aerobic fat power as Canova and Arcelli call it–by running at 90-100% MP is a key component of marathon training.
- The lactate threshold is determined by both lactate production and reuptake, and separate workouts and aspects of fitness modulate each of these factors.
- Runners with a higher proportion of fast twitch muscle fibers should emphasize different workouts than their “pure” slow twitch counterparts.
Aerobic fat power and long fast runs
When it comes to aerobic fat power, I haven’t seen any other coaching reference discuss the same rationale that Canova and Arcelli use. But from my physiology knowledge, it seems like quite a reasonable claim, and might explain why I’ve found long fast runs at 90-95% MP to be so effective for the athletes I coach.
I have some running energetics data that I’d like to look into to see whether I can find empirical data to back up this ‘maximize fat oxidation at 90-100% MP’ claim. I’ll update this post with a link once I do that.
Lactate threshold and fast vs. slow twitch
Canova and Arcelli’s focus on both sides of the lactate threshold equation—both its production and its reuptake—is an improvement, in my opinion at least, from traditional American-style training. Their perspective is borne out of a more careful consideration of how lactate is generated in the body, how it makes its way into and out of the bloodstream, and its ultimate fate.
Having one set of workouts that increase lactate production capabilities and another set of workouts that increase lactate reuptake in slow-twitch muscle fibers gives coaches a broader set of tools for boosting aerobic fitness, especially in more fast-twitch-like runners.
After reading this book, I’m more inclined to prescribe more uphill runs and fast progressive runs in athletes with a 3k/5k background, and less hesitant to use more traditional intervals at 8k-5k pace in runners coming “down” to the marathon from a trail or ultra background.
The evolution of Canova’s training approach over time 🔗
When comparing the marathon training approach outlined in this book with some of Renato Canova’s more contemporary writing, a few notable differences emerge:
First, there is no mention in this book of the special phase, which serves as a transitionary period between the fundamental phase and the specific phase and lasts 6-8 weeks. Here’s what I wrote about Canova’s special phase back in 2011 in my article Something New in Training:
The special period is focused on developing both the speed and endurance for the event, but never both at the same time. During the special period, which also lasts about two months, the short, middle, and long-distance specialists begin to diverge. All athletes try to maintain their mileage during the special period, so they are in a constant state of training, yet are not too tired for quality workouts. Athletes compete in races outside of their specialty; everyone but the marathoners competes in longer races, including cross-country. The marathoners compete at shorter distances, typically 10km, half marathon, or cross-country. Workouts from the fundamental period are not abandoned; they are added to the mix of workout options during the special period.
Something New in Training: The Methods of Renato Canova
The addition of this special phase might’ve been what Canova was referring to in his 2011 interview where he said that previously, the kind of training runners did to race the marathon now only means they are ready to start training for the marathon. Though it’s tough to say without seeing a full training schedule, the transition from fundamental training to marathon specific training does seem quite abrupt in the book–perhaps the special phase is a way to ease and improve that transition.
Also notably missing is the ‘something new’ that I described in that 2011 article, which was the evolution in thinking about mileage for more experienced runners, spurred by Italian compatriot Claudio Berardelli. In Canova’s later writing, there’s more emphasis on his belief that more experienced runners need less mileage, but more high-intensity running, since their “aerobic house” is already built–there’s no further benefit in continuing to punish the body’s mechanical integrity with the prodigiously-high volume referenced in this book. Do note that “lower” here still means ~90 miles a week, but that still pales in comparison to the 140+ mentioned in this book.
I wish this book had more details on how training progresses over time, both within a season and across an athlete’s career. The contrast in workout volumes between the fundamental phase and the specific phase suggest that there is a gradual, progressive increase in workout volume across the season: for example, the table of speeds above recommends 60-90 minutes progressive at 92-95% MP during the general phases (introductory + fundamental), and 36-45 km in this pace range during the specific phase.
These contrasting recommendations suggest there is a gradual increase in volume at 92-95% MP over the course of training, but the book doesn’t provide specifics on how often to run this pace (or any others), and how much to increase the volume over time.
Finally, though it’s mentioned briefly in Marathon Training - A Scientific Approach, Canova’s more recent writing puts much more emphasis on the shift in “modulation,” or day-to-day variability in mileage and intensity, during the specific phase.
Again, a day-to-day training schedule would have helped reveal whether Canova really does use more rest and recovery between marathon specific training sessions, but this is another point he specifically mentioned in his 2011 interview when discussing changes from this 1999 book–now, he says, runners sometimes need five or six days of easy recovery following major workouts before going for another long training session.
Renato Canova vs. Daniels and other typical American training approaches🔗
Marathon Training - A Scientific Approach frames the problem of training very similarly when compared with contemporary American training approaches[35]–it first outlines the major physiological contributors to performance, then outlines specific training sessions that improve each of those physiological aspects. Both Daniels’ Running Formula and my own book follow this structure. What’s interesting is where Canova and Arcelli diverge from the American style of training.
Physiologically-based training
American-style physiologically-based training involves a focus on 2-5 min intervals at VO2 max pace, and 3-8 minute repeats or 18-25 min continuous runs at but not exceeding lactate threshold.
In contrast, Canova and Arcelli de-emphasize very fast VO2 max-type intervals, preferring instead to use hill sprints to target the “central” component of the oxygen delivery system, and use the more marathon-relevant paces of 105-112% MP to target the oxygen uptake capabilities of the muscles.
Lactate threshold training
When it comes to training the lactate threshold, it’s long been taken as an almost sacred proposition that the lactate threshold must not be exceeded in Daniels-style cruise intervals.
Canova and Arcelli, instead, emphasize that you should be pushing your body modestly above the lactate threshold to “embarrass” the body into producing more enzymes for shuttling and reprocessing lactate. In this regard, their strategy is similar in its consequences to “CV pace” popularized by Tinman, and the “crest-load” pace promoted by John Kellogg, though not justified via the same critical power type reasoning.
Aerobic fat power training
As discussed above, the focus on improving aerobic fat power is an aspect of Canova and Arcelli’s physiologically-based approach that is virtually absent in popular American training approaches. In my own coaching, I’ve found that long fast runs at 90-95% MP are one of the most reliable kinds of workouts for improving marathon performance, and aerobic fat power might explain why.
Fast-twitch vs. slow-twitch marathoners
Canova and Arcelli’s consideration of fast-twitch marathoners differs somewhat from contemporary American takes. Steve Magness’ book, for example, has some great writing on differentiating training for fast-twitch versus slow-twitch runners. However, Canova and Arcelli are focusing on a very specific subtype of fast-twitch runners, namely those who have high levels of fast twitch oxidative fibers. Their criteria for being a fast-twitch runner include being successful at 5k and 10k on the track.
Usually Americans (including me!) think of this as indicative of not being a fast-twitch runner. But that’s because we’re thinking about the fast twitch glycolytic fibers, which contribute more strongly to 400, 800, and 1500 speed, and can’t use oxygen as effectively at submaximal speeds.
Canova and Arcelli’s approach for the fast-twitch athlete is very useful when working with runners who already have fast track PRs and are moving up to the marathon, but may not work as well for someone who was an 800 or 1500m specialist in high school or college who is transitioning to the marathon. These “true” fast-twitch runners probably do need to raise their VO2 max and lactate threshold, though perhaps through different training approaches than the true slow-twitch marathoner.
One final word on fast-twitch versus slow-twitch runners: occasionally I’ll see people who have never done higher-volume, aerobically-based training before naively assume they’re a fast-twitch runner because they specialized in shorter distances in high school, and their PRs get progressively worse at longer distances. But often this just ends up being an indicator of not enough training! If you are doing solid training, and you still find yourself thriving to a much greater extent in moderate-distance intervals (600-1600m repeats, for example), yet struggle in long, fast runs, you might be the kind of fast-twitch runner Canova is talking about.
Marathon-specific workouts
When it comes to marathon-specific workouts, Canova and Arcellis’ workouts look very different from American-style sessions, which involve long easy runs, an 8-10mi block of easy running followed by a block at MP, or short to moderate-length 100% MP repeats with long easy recovery in between. Classic examples of American-style marathon workouts include 22mi easy long runs, or 10mi easy + 10mi at MP, or 8mi easy + 3-3-3 mi at MP with 1mi easy recovery.
Canova and Arcelli’s approach is different. Aside from Canova and Arcelli’s “long run with long pickups,” most of the marathon-specific workouts recommended in Marathon Training - A Scientific Approach are not done following long blocks of easy running, and instead involve either continuous runs at fractionally slower than marathon pace (90-98% MP), or involve moderate to long reps at 100-105% MP with the recovery at a still-quite-fast 93 to 97% MP.
The primary physiological goals that Canova and Arcelli provide for these two workout types are (1) increasing aerobic fat power, and (2) improving lactate reuptake.
To the extent that you believe Canova and Arcelli’s physiological arguments, it’s easy to see why American-style workouts would be less effective in both regards: they do not involve running at the speeds that maximize fat oxidation, and the recovery in between MP reps is too slow to “embarrass” the body’s lactate reuptake capabilities.
Adapting Renato Canova’s marathon workouts for non-elite runners 🔗
Even for fairly high-level runners, some of the workouts that Canova and Arcelli recommend in this book are quite demanding–and that’s before considering the training volumes of up to 120-180 miles a week mentioned in Chapter 4! I wouldn’t recommend adopting these workouts whole-cloth without modifications unless you really know what you’re doing.
I’ve had success using the following techniques to adopt Canova’s workouts with the athletes I work with as a coach, and in my own training:
- Reduce workout volume: When I compare notes against the table of marathon paces above, I find that I tend to use only 60-80% of the total volume that Canova and Arcelli recommend for the various marathon speeds.
- Example: 4-5 x 2 km at 110% MP, instead of 5-6 x 2 km
- Introduce workouts at much lower volume: Not only do I use lower total workout volumes; I also build up to the max workout volume very gradually, both in terms of the rep distance and in terms of the workout volume.
- Example: Starting with 6 x 1 km at 110% MP, building over the training cycle to 4-5 x 2 km
- Increase recovery and “modulation” surrounding big workouts: When incorporating long, fast runs and workouts at close to marathon pace, lower-mileage runners in particular do better when they take several shorter, easy days following the workout. During the specific phase, I’m almost totally unconcerned with overall mileage–the main goal is to be fresh for the marathon-specific workouts and to recover well afterwards.
- Example:
- Sat: 60min easy
- Sun: 18mi at 95% MP
- Mon: Off or 20-30min very easy
- Tue: 45-55min easy
- Wed: 60-70min easy to moderate with light pickups
- Thu: 60min easy + strides
- Fri: [next workout]
- Example:
- Use short blocks of volume instead of high weekly mileage. Even for non-elite runners, the marathon is still 26.2 miles long–as a consequence, it’s beneficial to do some workouts on “tired legs,” as if you were in the middle of a high-volume block. So, you can stack 2-3 days of higher volume together, then take several shorter, easier, lower volume days afterwards.
- Example:
- Thu: 8mi easy
- Fri: AM: 10mi easy to moderate with light pickups; PM: 6mi easy
- Sat: 20mi at 90% MP
- Sun: Off
- Mon: 5-6mi easy
- Tue: 8-9mi easy with light pickups
- Wed: 13mi with 4x1.5mi at 105% MP
- Weekly total is 70-72 miles, but Fri + Sat average 18mi per day.
- Example:
- Focus on 90% MP and 110% MP workouts initially. In Canova’s more recent parlance, these are “special” speeds, which provide support for later marathon-specific work at 95-105% MP. Long fast runs at 90% of MP seem to go pretty well with less-experienced, lower-mileage runners, and 110% MP corresponds to around 10k-ish pace, so those also tend to be an easy transition.
- More advanced athletes with higher mileage handle marathon specific work better. The workouts that tend to require more mileage and experience are long fast runs at 95% MP, marathon-specific repeats at 100-103% MP, and fast continuous runs at 105% MP; they can often wait until a later training cycle to be introduced in full force.
Conclusion 🔗
Hopefully you’ve found this review and discussion helpful. If you did, you can support my work by sharing this article with friends or posting about it on social media.
I also have a book of my own if you want to see how I combine physiology with modern coaching approaches. If you want to get notified when I publish new content on Running Writings, be sure to sign up for my email newsletter below.
Recommended reading for marathon training 🔗
I wish I could link you to a place to buy Canova’s book–but of course the entire motivation for this article is that the book is nearly impossible to find! Instead, here are some additional books and articles that I can recommend for learning more about high-level marathoning in general, and Canova’s training in particular.
- Running Faster from the 5k to the Marathon - Brad Hudson is widely cited by many people as the most Canova-influenced American coach, so it’s a good read if you want to see how another marathon coach puts Canova’s principle into action.
- Train Hard, Win Easy and More Fire by Toby Tanser - These books are a great window into what early-career training looks like for the top Kenyan runners who end up working with coaches like Renato Canova. The training Toby describes is much more freewheeling and feeling-based, and while you won’t find training plans in the books, you will get a better idea of the Kenyan mentality about training and racing.
- Something New in Training - My 2011 article summarizing Canova’s training philosophy from 800m up to the marathon. Includes a discussion of the “special phase,” which is not mentioned in Marathon Training - A Scientific Approach.
- Daniels’ Running Formula - The canonical “American-style” physiologically-based training book. The chapters on targeting specific physiological adaptations in training are an interesting contrast with Canova and Arcelli’s approach.
- Elite Marathoning with Renato Canova: The Training of Moses Mosop and Abel Kirui - My 2012 article with training calendars from two world-class Canova-coached marathoners.
- Gelindo Bordin’s training - Seeing the real training of an athlete trained under the Canova/Arcelli approach is very helpful for getting a better idea of how all of the workouts in the book come together. I plan on doing a write-up on Bordin’s training and will update this post with a link when I’ve got that finished.
- The Science of Running - Steve Magness’ book features the most comprehensive treatment of training differentiation for fast-twitch and slow-twitch distance runners that you’ll find in the coaching literature. It’s great for its modern take on physiology, too.
Some of the links above are affiliate links which help support my work.
Footnotes 🔗
[1] Like this one. Click the footnote number to the left of the note to jump back up to your spot the main text.
[2] mmol/L is a chemistry unit that measures the number of molecules in a given volume of liquid; in this case it refers to how many molecules of lactate there are in one liter of blood.
[3] For technical reasons, physiologists hate the term “anaerobic threshold”; it’s more often called the “lactate threshold” in modern physiology research. Physiologists likewise dislike referring to intervals or speeds as “lactic” or “anaerobic.” Regardless, I use the terms Canova and Arcelli use in the book in this article for consistency.
[4] One note on terminology: though the book sometimes talks about “percentages of speed,” it’s pretty clear from the paces in the book and the context of the rest of Canova’s work that this actually means “percentages of pace”–from a strict mathematical perspective percent of speed and percent of pace are not the same thing. See the linked blog post in the previous sentence for details on why. I use the term “percent of pace” in this article for consistency, and the abbreviation “% MP” for percent of marathon pace. Unless otherwise noted, when I use “speed” I mean it only in the general sense of “running faster.” All percentages are given in percent of pace.
[5] For reference, Heartbreak Hill at Boston is a 3.3% grade; the steepest allowable grade on US interstates is 6%; and the Pike’s Peak Ascent has an average grade of 11%. A 15% grade is really steep!
[6] There are differing modern opinions on the mechanism behind hill sprints; many coaches think the benefits are purely neuromuscular, with hill sprints essentially functioning like running-specific plyometrics.
[7] Presumably this would involve determining the heart rate at the anaerobic threshold via a field test (Chapter 3), then calculating the heart rate equivalent of 97-105% of anaerobic threshold pace.
[8] As with the methodologies for taking lactate samples, I lay no claims to the validity of the Conconi test–I’m just reporting what’s in the book. Research indicates that the Conconi test may not be an accurate method for determining lactate threshold, and not all runners even show a deviation from linearity in heart rate during an incremental test. I actually happen to be one of these linear-only runners!
[9] Be aware that the Conconi test as-described is a constant progression of speed, not pace.
[10] I created this and the following plots using data extracted from the plot in the book.
[11] The blue line is a generalized additive model with cubic regression splines, if you care to know the details. My range of plausible values for Vd come from sweeping 80 to 99% confidence intervals on where the derivative of the smooth deviates from the slope of the linear fit, which is fit to only data below 17 km/hr ("eyeballing it").
[12] This lactate testing protocol is not necessarily “ideal” or the one I would recommend; it’s just the one that’s in the book.
[13] This 10sec/km pace change is contrary to the 2.5-5 sec/km pace changes recommendations made in the text.
[14] I suspect this situation has changed since 1999, though I don’t have first-hand experience with portable metabolic analyzers.
[15] Percentages for this workout are based off Curatolo’s 2:30:33 in that race, which is only 17 seconds slower than her all-time PR.
[16] Percentages for this workout are based off Milesi’s PR of 2:11:58, which he ran in Berlin the following spring. Wolfram Alpha tells me that it was cool but windy in Athens for the World Cup race; his time in that race was 2:13:23.
[17] Percentages for this workout are based off Ornella’s 2:31:30 at the Rome Marathon the following spring; the World Championships race was actually 400m short because of a course error, and Ornella’s time in that race (2:30:11) is very close to her Rome time if you adjust for the short course. However, in the book Canova talks about this “extensive” work as being more like 98% of MP, so it’s more likely this was intended to be more like 98% MP.
[18] The main text says 5:35 for 2k, which doesn’t make sense. Table 9 says 5:25, so I corrected it for these calculations.
[19] Many of the terms for paces and workouts in this book are inconsistent, particularly with regards to “aerobic power” and “aerobic endurance.” Be extra cautious when trying to interpret anything dealing with these terms, since each one has three or four different variants which seem to refer to the same thing (e.g. aerobic endurance, aerobic lipidic power, fat power).
[20] Adapted from Table 7 in the book, which appears in Chapter 3.
[21] Adapted from Table 9 in the book.
[22] Also called “interval runs with long/medium/short variations.”
[23] The heart rate intensity here is based on my own estimate using the description of long uphill runs from earlier in Chapter 4.
[24] Table 10 in the book has a significant misprint: The “loads” and “examples” columns are duplicates of Table 9, which is the table about increasing oxygen uptake. I’ve filled in the “Distance and Pace” and “Example” columns as best I can using the information and examples given earlier in the chapter. Some workouts are not described in enough detail to reconstruct.
[25] The printing error described above caused the example workouts in Table 9 to be repeated in Table 10; sometimes there was no example session from earlier in the book to drop in, hence the N/As.
[26] There are no guidelines anywhere in the book on the length or intensity for long runs through hills.
[27] Some of the example workouts are given using a hypothetical 2:08 male marathoner, and some are given using a hypothetical 2:28 female marathoner. I’ve translated all paces to % MP to facilitate easier comparisons. There doesn’t seem to be any differences for males vs. females in prescribing the workout volumes.
[28] The book says 500m recovery in 1:55 for a 2:28 marathoner, which is 91% MP and doesn’t seem right. I left this at 97% MP which is what is suggested in the “Loads” column of the table.
[29] The book says 98-100% MP, but this seems to be a mistake given the example workouts detailed earlier in the chapter, which suggest 96-98%.
[30] Readers familiar with Canova’s other writing might notice that the “special” phase is missing; it usually follows the fundamental phase and precedes the specific phase. See the discussion section for more thoughts on why this phase is missing.
[31] No explicit guidelines or examples are given on any of these exercises, though the uphill sprints sound identical to those described earlier.
[32] Table 7 in Chapter 3 suggests “slow runs” should be done at 80-85% MP.
[33] Again, Table 7 indicates medium runs should be done at 90-92% MP.
[34] Indeed, the term “aerobic efficiency” is used in Table 10, alongside the phrase “aerobic extensive power” which might mean something different from “aerobic power.”
[35] Here I’m specifically referring to the kind of training you see discussed online or in commonly-recommended training books. I suspect many top American coaches and elites already employ long fast runs at 90-95% MP, even if it isn’t justified via targeting aerobic fat power.
Related articles

The Keys to Marathon Training: Modern changes to Renato Canova’s elite marathon training methods

Preparing for championship races with Renato Canova: Caleb Ndiku's training before the Commonwealth Games 5k

Championship Racing with Renato Canova: Caleb Ndiku's indoor 3k training

Elite Marathoning with Renato Canova: The Training of Moses Mosop and Abel Kirui

Preparing for championship races with Renato Canova: training for the 1500, 5k, and 10k

Great article! Congratulations on your dissertation!