What Happened in Montréal
The UCI’s race report gives the course: 184 riders from 61 nations started a 273.4 km race with 3,803 m of climbing: a loop of about 100 km, then 12 laps of the 13.4 km Mont-Royal circuit. Each lap climbed the Voie Camillien-Houde, 1.67 km at 7.4% with a maximum of 10%.
According to the UCI, a 12-man group came back together for the final 35 km, and McNulty went clear alone with 33 km to go. Reuters and Escape Collective put the attack at 32 km to go; this article uses the UCI’s figure. He led by a minute before the last two laps and by 53 seconds at the start of the final lap. Past Camillien-Houde, with 10 km left, the gap was down to 10 seconds.
It held. McNulty won in 6:17:04 (Reuters and The Canadian Press), 13 seconds ahead of Michael Matthews, with Mathieu van der Poel third (UCI). The Canadian Press lists Matthews, van der Poel and Quinn Simmons together at 13 seconds. Reuters reported the win as the end of a 33-year wait for a US men’s road world title. Distance and time work out to an average of about 43.5 km/h (our arithmetic). Of the last climb, McNulty said, as quoted by the UCI: “I was just dying, but I kept pushing.”
What this article does not have is his power file. We found no race power data from McNulty in any source we could read, so nothing below estimates his watts, kilojoules or kJ/kg. What the research can explain is why the last 33 km of a 273 km race is a different problem from 33 km ridden fresh.
What Durability Means
Maunder and colleagues (2021) introduced the term to describe when, and by how much, an athlete’s physiological profile deteriorates during prolonged exercise. Cycling studies measure it simply: compare your best power for a fixed duration when fresh with the same effort after a set amount of work. The work is often given in kilojoules per kilogram of body mass (kJ/kg). Work in kilojoules is watts × seconds ÷ 1,000, and dividing by body mass lets riders of different sizes be compared.
Three findings recur across the papers read for this article.
- Even elite riders lose power. Ørtenblad et al. (2024) had 12 elite men ride four hours of intermittent work while taking 100 g of carbohydrate per hour. Their 6-minute time-trial power fell from 415 to 375 W, and their 6-second sprint peak by 6%.
- Riders differ widely. In that study the change in 6-minute power ranged from −31% to +1%. In the studies reviewed by Jones and Kirby (2025), critical power fell by about 10% on average after two hours of heavy exercise, but by anything from 1% to 33% between individuals.
- Fresh numbers predicted it poorly. Neither Ørtenblad et al. nor Barsumyan et al. (2025), who tested 20 well-trained amateurs, found a strong link between the size of the drop and VO2max, FTP or other measures taken fresh. Barsumyan’s group concluded that durability is independent of the traditional measures; Jones and Kirby treat it as an additional factor in endurance performance. Both papers note an exception: Spragg et al. (2023) found strong links between fresh lab measures and the fall in critical power, in riders Ørtenblad et al. describe as elite under-23s.
Best power lost from fresh to fatigued, by effort length
Why Intensity Matters More Than Kilojoules
The finding that matters most for reading your own data comes from the 2025 systematic review by Sánchez-Jiménez and colleagues in the European Journal of Applied Physiology, which covered 21 studies. Across them, power fell by 10–20% after 2.5–15 kJ/kg of prior work above critical power (CP), while similar or larger volumes below CP typically cost less than 5%. Kilojoules were the most common fatigue measure, and the authors conclude that kJ alone may be insufficient, because it ignores how hard the work was done.
Mateo-March et al. (2024) show the effect in 17 male professionals over a season. After work done above CP, their critical power fell by 2.2%, 6.1% and 16.2% at 2.5, 5.0 and 7.5 kJ/kg. After the same amounts of work below CP, there was no significant change.
Fall in critical power after the same work, done above or below CP
After 2.5 kJ/kg
After 5.0 kJ/kg
After 7.5 kJ/kg
Level matters as well. In Leo et al. (2024), as summarised in the review, under-23 riders lost power in every effort of a minute or longer after 2.5–5.0 kJ/kg above CP, while ProTeam and WorldTour riders only showed significant drops after 5.0–7.5 kJ/kg. In professional field data (Mateo-March et al. 2022), declines grew from 1.6–3.0% after 15 kJ/kg to 6.0–9.7% after 45 kJ/kg.
The review names races with cobbles or repeated short climbs, such as the Tour of Flanders, as demanding repeated high-intensity efforts. Montréal put 12 ascents of Camillien-Houde into its final 160 km. How much of McNulty’s work was above his own CP is unknown without his power file. For what critical power is and how it is estimated, see our critical power explainer.
How to Measure Your Own Durability
Option 1: a two-session test
Barsumyan et al. (2025) built their protocol for amateurs, because the professional tests, with rides of 3–4 hours or around 40 kJ/kg, are impractical for them. Twenty well-trained men (mean age 34.6, 74.4 kg, about 8 hours of training a week) did it at home on their own bikes and smart trainers, on two consecutive weekends.
Durability test protocol card
Session 1: fresh
- 120-min standardized warm-up
- 25-min all-out effort
- 340 min easy (no harder than 2 out of 10 on the Borg scale)
- 420-min all-out effort
Session 2: fatigued (the following weekend)
- 120 min at 70% of your fresh 20-min power
- 2Continue at 80% of fresh 20-min power until 1,000 kJ in total (warm-up included)
- 35-min all-out effort, straight away
- 440 min easy (no harder than 2 out of 10)
- 520-min all-out effort
- Same conditions both days: breakfast of about 2 g/kg carbohydrate and 800 mL water one hour before
- 60 g/h carbohydrate (1:1 glucose to fructose) during both tests; no caffeine during the fatigue ride
- No caffeine, alcohol or hard training in the 24 h before; efforts held at 80–100 rpm
- Own bike on a calibrated smart trainer, with fans, in a 15–20 °C room
5-min power, study mean
4.3 → 3.8 W/kg
−10.8%
20-min power, study mean
3.5 → 3.2 W/kg
−10.1%
Worked example (a hypothetical rider, not study data)
A rider whose fresh 20-minute power is 250 W rides the warm-up at 175 W (70%): 20 minutes is 210 kJ. The remaining 790 kJ at 200 W (80%) takes about 66 minutes, so the 5-minute effort starts roughly 86 minutes into the ride. If the fatigued 20-minute effort then averages 230 W, the drop is 8%. At 75 kg, 1,000 kJ is about 13.3 kJ/kg; a lighter rider doing the same 1,000 kJ carries a bigger relative load, so keep the target fixed each time you retest.
Option 2: your existing ride files
The field studies use training and race files instead. They group each rider’s best efforts by how much work came before them; Mateo-March et al. (2022) used 0, 15, 25, 35 and 45 kJ/kg. A simple version for your own data:
- Pick a threshold your long rides pass, such as 15 kJ/kg (1,125 kJ for a 75 kg rider).
- Find your best 5- and 20-minute power before that point, and again after it.
- Note how the earlier work was done, such as time above CP or FTP, not just the kJ total, because the review shows hard work costs more than steady work.
- Only compare rides where you actually went hard late. A missing effort is not a measured loss.
Read the two curves side by side, as in how to read your power curve: the gap between them is your durability at that load.
How to Train It
The evidence on training durability is thinner than the evidence on measuring it. Jones and Kirby (2025) note that data are scant, then list what the available evidence suggests might help:
- Consistent training and large volumes accumulated over several years.
- Regular long sessions, especially ones with efforts at race pace or above, or with intensity rising as fatigue builds.
- Heavy strength and plyometric training. In one study they cite (Rønnestad et al.), elite cyclists who added heavy strength training twice a week for 12 weeks rode a better 5-minute time trial at the end of three hours of moderate riding than a control group that only did endurance training.
The Sánchez-Jiménez review argues for building tolerance to high-intensity work in particular, and points to a session from Spragg et al.: 20 minutes at 50–70% of CP followed by five 8-minute intervals at 105–110% of CP. Spragg’s group used it with under-23 Continental riders, and the review suggests it could mimic Grand Tour stage demands. For the above-threshold side, see Ben Healy’s 42 km solo.
Fuelling is part of the picture. In one study reviewed by Jones and Kirby (Clark et al., in non-elite participants), 60 g/h of carbohydrate during two hours of heavy exercise led to a smaller fall in CP than a placebo. Even at 100 g/h, Ørtenblad et al. estimated that about 270 g of carbohydrate still came from the riders’ own glycogen over four hours; they did not measure glycogen directly. The Sánchez-Jiménez review, citing Peeters et al. (2025), adds that carbohydrate intake in the days before a durability test, not only during it, should be controlled. For intake on the bike, see on-bike fueling.
Then retest. Ørtenblad et al. cite Matomäki et al. (2023): in untrained people, ten weeks of training changed VO2max and power at the first lactate threshold, but those changes were not linked to changes in durability. Better fresh numbers do not guarantee better late-race numbers; a repeat of the same test is how you find out.
What to Take From McNulty’s Ride
McNulty’s winning move came after 240 km of racing. The studies above show why late-race power is its own ability: the loss differs widely between riders, fresh lab numbers predicted it poorly in most of these studies, and work above CP costs far more than steady work. For an amateur, the practical step is to measure your drop, then train with hard efforts placed late in long rides.
Sources
- Race facts: UCI, “2026 UCI Road World Championships: McNulty caps off memorable Montréal week” (uci.org, 28 September 2026); Reuters (Rory Carroll and Vincent Daheron), “McNulty wins road world title to end US men’s 33-year drought”, via KSL, 27 September 2026; The Canadian Press, “American Brandon McNulty claims men’s road race world title in Montreal”, via CP24, 27 September 2026; Escape Collective, “Worlds: McNulty goes solo to secure a rainbow jersey in Montreal”, 27 September 2026.
- Research: Sánchez-Jiménez JL et al., “Is intensity the most important factor in determining the amount of prior work accumulated that affects cyclists’ acute durability? A systematic review”, European Journal of Applied Physiology 126:163–179 (online 4 July 2025), doi:10.1007/s00421-025-05885-0. Barsumyan A et al., “Durability as an independent parameter of endurance performance in cycling”, BMC Sports Science, Medicine and Rehabilitation 17:192 (July 2025), doi:10.1186/s13102-025-01238-8. Ørtenblad N et al., “Substrate utilization and durability during prolonged intermittent exercise in elite road cyclists”, European Journal of Applied Physiology 124:2193–2205 (2024), doi:10.1007/s00421-024-05437-y. Mateo-March M et al., “Is all work the same? Performance after accumulated work of differing intensities in male professional cyclists”, Journal of Science and Medicine in Sport 27:430–434 (2024), doi:10.1016/j.jsams.2024.03.005 (abstract). Jones AM, Kirby BS, “Physiological resilience: what is it and how might it be trained?”, Scandinavian Journal of Medicine & Science in Sports 35:e70032 (March 2025), doi:10.1111/sms.70032.
- Studies cited through these papers: Maunder et al. (2021), Leo et al. (2024), Mateo-March et al. (2022), Sánchez-Jiménez et al. (2023), Spragg et al. (2023) and Peeters et al. (2025), as summarised in the Sánchez-Jiménez review (Spragg et al. 2023 also as cited by Barsumyan et al. and Ørtenblad et al.); Rønnestad et al. and Clark et al., as summarised by Jones and Kirby; Matomäki et al. (2023), as cited by Ørtenblad et al.
The worked example uses invented round numbers.