Intensity distribution: how much easy, how much hard

The question is how much, not which

The argument about easy against hard usually starts in the wrong place. It is set up as a choice between two methods, as though you had to pick a side. Nobody who trains well does that. The hard sessions and the easy hours each solve a different problem, and why they do is covered separately.

The open question is the distribution: how much of the time should be easy, how much should be hard, and how well we actually know.

The distribution is not a law of nature. It is a choice about control, and the choice looks different for someone training nineteen hours a week and someone training eight.

What the professionals actually do

There are measurements now, not only anecdotes. In the best available study, 32 WorldTour professionals were followed through 1 696 training sessions in the ten weeks before their season started. Measured by power, the men spent 70.4 ± 6.2 % of the time at low intensity, 25.4 ± 5.1 % in the middle zone and 4.2 ± 1.8 % at high intensity. The women were at 59.7 / 32.8 / 7.4 % (Mateo-March et al. 2026).

More than 90 % of the riders followed a pyramidal distribution, regardless of sex (Mateo-March et al. 2026). Not polarised.

The shape is not new. When eight professional cyclists were measured in racing, across 22 stages of the Tour de France, their time fell at 70, 23 and 7 % across the three zones (Lucía et al. 1999). That is racing load rather than training, but the pattern is the same, 25 years earlier.

What separates most is the volume: 19.1 ± 2.7 hours a week for the men (Mateo-March et al. 2026). That is the part of professional training an amateur cannot copy.

What "80/20" actually counts

The rule everyone quotes comes from Seiler's review of training diaries. It says that about 80 % of the sessions are done at low intensity — not 80 % of the hours. And it applies to a particular group: athletes training ten to thirteen times a week (Seiler 2010).

That distinction matters. Someone training five times a week does not have ten sessions to distribute, and a share of sessions is not a share of time. The numbers in the section above are shares of time, and they are lower.

The authors have asked the question themselves. In a review written for coaches, Seiler and Tønnessen ask whether the 80/20 pattern suits recreational athletes training four to five times a week at all, and answer that there are almost no published data addressing it (Seiler & Tønnessen 2009).

Polarised or pyramidal

The two models describe where the hard part sits. Polarised means plenty of easy work and a little very hard work, with little in between. Pyramidal means the volume tapers gradually as intensity rises, with more work in the middle zone.

The largest analysis in the field pooled individual data from 13 studies and 348 athletes. It found no difference between polarised and pyramidal, for VO2max or for time-trial performance (Rosenblat et al. 2025). The same analysis found the response depends on level: competitive athletes may get more from polarised, recreational athletes more from pyramidal.

Another meta-analysis finds an advantage for polarised on VO2peak, but only in interventions shorter than twelve weeks and only in well-trained athletes (Silva Oliveira et al. 2024). And when you look at what athletes actually do, the pyramidal pattern is the most common in the literature (Stöggl & Sperlich 2015).

Your level therefore matters more than the name of the model: the same analysis finds the response depends on how well trained the athlete is (Rosenblat et al. 2025), while what recurs in the descriptive material is that the bulk of the training sits low (Stöggl & Sperlich 2015).

The middle zone — and what "threshold training" means in a study

We have to be precise here, because one term means two different things.

The difference decides how you read the findings. When four training approaches were compared over nine weeks in 48 well-trained athletes, threshold work alone produced no measurable gain in the endurance variables, apart from a small improvement in work economy (Stöggl & Sperlich 2014). And with the hard share held equal, more of the volume below the first threshold produced greater improvement than more of the volume in the middle zone (Esteve-Lanao et al. 2007).

That is an argument against letting the middle zone eat the easy hours. It is not an argument against sweet spot sessions with a purpose. Our base period is built deliberately on sweet spot alongside zone 2, and the difference lies in dose and placement: a planned sweet spot session is not the same as every ride drifting up into the middle zone because it feels good.

That last mistake is the common one. Seiler and Tønnessen describe it this way: it is hard to keep recreational athletes training 45–60 minutes three to five days a week from accumulating a lot of time around threshold (Seiler & Tønnessen 2009). The sessions become too hard to give volume, and too easy to give a real stimulus.

What this means when you have 6–12 hours

This is where the professional data stops being relevant, and a different set of studies takes over.

First, a warning about the numbers. The measuring instrument changes the answer by itself. In 15 well-trained cyclists, across the same ten weeks and the same rides, the share of time at low intensity came out at 44.9 % measured by perceived exertion, 86.8 % by heart rate and 79.5 % by power (Sanders et al. 2017). If you compare your heart-rate zones with a professional's power zones, you are comparing two different quantities. The number in the next paragraph and the number further up resemble each other, but they were measured different ways and should not be read as the same thing.

With that caveat: recreational athletes left to steer for themselves land on a pyramidal pattern. When 48 recreational runners were followed for twelve weeks with GPS and a heart-rate monitor, their training fell at 69.1 % low, 17.3 % moderate and 13.6 % high intensity (Bonato et al. 2026).

Polarised training is not free for everyone either. In a study of 42 recreational runners, more responded to low-intensity training than to polarised training or to pure high-intensity blocks (Zinner et al. 2018).

And volume itself is not the lever it is often made out to be. In a review of 41 studies and 797 cyclists, the authors found no association between weekly or total training volume and change in VO2max or time-trial performance (Cove et al. 2025).

The sum is simple: do not copy the distribution of someone training nineteen hours. Choose the distribution you can carry out week after week, and make sure the bulk of it is easy.

The sessions the distribution is made of

The distribution is not only numbers. It is sessions. Zone 2 carries most of the volume and is covered separately, while the three quality types we use each have their own job. The frames below are our own, taken from our training philosophy.

The numbers for each zone are in the zone article. What the sessions look like across a week is in the programmes.

How we do it

Our structure is 1–2 quality sessions a week, with zone 2 around them. That is what the programmes are built on, and the rest of the week's load follows from it.

We steer the total load with the zone 2 volume rather than with more intervals. That does not contradict the finding above: the point is not that hours in themselves produce progress, but that easy hours are the safest way to raise the load when it has to go up.

What changes through the year is the dose in the quality sessions, not which types of session you do. How the year is put together in blocks is covered in the periodisation article.

If you want a ready-made week, they are in the programmes. The numbers for each zone are in the zone article.

What we do not know

The weakest link in this article is not the argument between models. It is the transfer.

The numbers further up come from one team, in ten weeks without racing, with zones set by critical power. They do not describe a whole year, and they do not describe you. For athletes at your volume the picture is thinner: in a systematic review of trained cyclists training 7.5–11.7 hours a week, the authors conclude that no evidence is currently available favouring a specific periodisation model over eight to twelve weeks (Galán-Rioja et al. 2023).

Add that there are almost no published data on whether the 80/20 pattern applies to recreational athletes (Seiler & Tønnessen 2009), and that the measuring instrument alone can move the share at low intensity by more than 40 percentage points (Sanders et al. 2017).

Sources

1. Mateo-March M, Barranco-Gil D, Muriel X, Pallarés JG, Valenzuela PL (2026). Biology of Sport 43(4):933–940. doi:10.5114/biolsport.2026.156234
2. Lucía A, Hoyos J, Carvajal A, Chicharro JL (1999). Int J Sports Med 20(3):167–172. doi:10.1055/s-1999-970284
3. Seiler S (2010). Int J Sports Physiol Perform 5(3):276–291. doi:10.1123/ijspp.5.3.276
4. Seiler S, Tønnessen E (2009). Sportscience 13:32–53. sportsci.org/2009/ss.htm
5. Rosenblat MA, Watt JA, Arnold JI, et al. (2025). Sports Medicine 55(3):655–673. doi:10.1007/s40279-024-02149-3
6. Silva Oliveira P, Boppre G, Fonseca H (2024). Sports Medicine 54(8):2071–2095. doi:10.1007/s40279-024-02034-z
7. Stöggl TL, Sperlich B (2015). Frontiers in Physiology 6:295. doi:10.3389/fphys.2015.00295
8. Stöggl T, Sperlich B (2014). Frontiers in Physiology 5:33. doi:10.3389/fphys.2014.00033
9. Esteve-Lanao J, Foster C, Seiler S, Lucia A (2007). J Strength Cond Res 21(3):943–949. doi:10.1519/R-19725.1
10. Sanders D, Myers T, Akubat I (2017). Int J Sports Physiol Perform 12(9):1232–1237. doi:10.1123/ijspp.2016-0523
11. Bonato M, et al. (2026). Int J Sports Physiol Perform 21(7):783–791. doi:10.1123/ijspp.2025-0348
12. Zinner C, Schäfer Olstad D, Sperlich B (2018). Med Sci Sports Exerc 50(8):1641–1648. doi:10.1249/MSS.0000000000001599
13. Cove B, Chalmers S, Nelson MJ, Anderson M, Bennett H (2025). J Sci Med Sport 28(5):423–434. doi:10.1016/j.jsams.2024.12.005
14. Galán-Rioja MÁ, Gonzalez-Ravé JM, González-Mohíno F, Seiler S (2023). Int J Sports Physiol Perform 18(2):112–122. doi:10.1123/ijspp.2022-0302