Durability Training: How to Train (and Measure) the Endurance That Wins Late in the Race

In a Tour de France mountain stage, twenty riders may crest the first climb together. Ten make it over the second climb. Five make it over the third. The riders who get dropped do not have lower VO₂max than the riders who stay on. Their fresh-state thresholds, on paper, look identical. What separates them is durability: how well their physiology holds up after hours of preceding work.

Durability is the single biggest gap in the way endurance athletes are profiled and the way training plans are prescribed. Lab tests measure everything in a fresh state, by design, but races are won and lost in the final 10 to 20 percent, when fresh-state numbers no longer apply. This page will cover what physiological durability is, why it matters, the three physiological systems that determine it, and how Tymewear measures it in the field.

What Durability Is, in One Sentence

Durability is the resistance of your physiological thresholds and capacities to drift downward during prolonged exercise.

A highly durable athlete still has roughly their fresh-state power at VT1, their fresh-state running pace at lactate threshold, and their fresh-state critical power even after several hours of preceding work. A less durable athlete, even with the same fresh-state numbers, sees those markers fall, sometimes by 10 percent or more, after a few hours of sub-maximal effort.

The term was formally introduced into the sports-science literature by Maunder and colleagues in 2021, who argued that traditional fresh-state physiological profiling (VO₂max, VT1, critical power) does not capture how an athlete performs late in a race, and that durability should be treated as a distinct dimension of endurance performance.

Why Fresh-State Lab Numbers Don't Tell the Whole Story

Two athletes can have the same VO₂max, the same VT1 power, and the same critical power, and still have very different race outcomes over a long event. Maunder describes seeing athletes with nearly identical fresh-state profiles produce wildly different long-event performances, and athletes who improved year-over-year in races without showing big shifts in their lab numbers. The variable that explains those gaps is durability.

The clearest experimental confirmation comes from cycling. After roughly 2.5 hours at 90 percent of fresh VT1, the SPRINZ research group has reported VT1 power drops ranging from 1 watt to 45 watts across individuals, with a mean drop of about 10 percent. The drop is non-linear: it accelerates early, then stabilizes. Athletes with poor durability also suffer larger declines in their ability to perform at severe intensities later in a bout, which is what races usually demand at the finish.

This is why a "fresh threshold power number + a generic training plan" approach systematically under prepares athletes for long events. The threshold the athlete trains around is not the threshold they will be racing on at hour three.

The Three Physiological Systems Behind Durability

Durability is multidimensional. Physiological signals measured during prolonged exercise decouple from external workload in distinct ways, and each pattern tells you something different about how the body is fatiguing. Three systems carry the signal:

1. The cardiovascular system, indexed by heart rate.
2. The metabolic and pulmonary system, indexed by minute ventilation (V̇E) and substrate use (fat and carbohydrate oxidation).
3. The neuromuscular system, indexed by breathing rate (BR, also written as respiratory frequency, fR).

These three systems each erode at their own rate during a long session, and the relative contribution differs between athletes. Reading all three together, rather than relying on heart rate alone, is the core of durable-state monitoring.

1. The Cardiovascular System: Heart Rate and Cardiac Drift

Heart rate at a fixed power or pace rises during prolonged exercise. This is cardiac drift, and it is the most familiar form of physiological decoupling. The mechanisms are well established: plasma volume falls, core temperature rises, stroke volume drops, and heart rate rises to compensate.

In the Rothschild 2025 dataset (51 trained cyclists, 85 measurement time points), HR drift at a matched power was the strongest single predictor of how much VT1 power had been lost during the prolonged bout. After 150 to 180 minutes of cycling, HR at VT1 was significantly higher than fresh state (147 vs. 142 b/min), even though VT1 power had dropped.

This sets up a serious problem for HR-based pacing. A fixed HR ceiling, like the popular "stay below 135 bpm for Zone 2," fails as the session progresses because the threshold has moved while heart rate has drifted upward. Maunder makes the applied point explicitly: an athlete using a fixed HR ceiling on a long run is likely under-training late in the session, because they can tolerate some HR drift and still be physiologically below their threshold. 

The compounding factor is that HR-based zones are unreliable in a fresh state too. Meyer and colleagues showed that "85 percent of HRmax" placed 36 trained cyclists anywhere from 87 to 116 percent of their individual anaerobic threshold. Adding Rothschild's roughly 10 percent threshold shift on top of that fresh-state variability, late in a long ride a fixed HR zone can be off by almost 26% from the athlete's actual current threshold.

HR drift is real, it is informative, and it correlates strongly with durability. But heart rate alone cannot tell you whether the drift is benign (plasma volume, thermoregulation) or deleterious (substrate depletion, genuine threshold erosion). For that, you need to read the other two systems alongside it.

2. The Metabolic and Pulmonary System: Ventilation and Substrate Shift

Minute ventilation (V̇E) is the volume of air moved per minute. It is tightly coupled to CO₂ production, which is in turn tightly coupled to metabolic demand and substrate use. Where heart rate is contaminated by drift, V̇E tracks genuine metabolic cost.

The signature finding for V̇E during prolonged moderate-intensity exercise is that it stays remarkably stable. Stevenson 2024 reported that during a 2-hour bout at 90% of VT1, V̇E did not significantly change, and that V̇E at VT1 was stable before versus after the 2 hour bout. Rothschild 2025 likewise found that V̇E drift was not a significant predictor of durability loss which is informative, not a null finding:

V̇E is the signal that does not get fooled by cardiovascular noise.

What V̇E does capture is the substrate shift toward fat that develops during prolonged work. In Jaén-Carrillo's 23 trained trail runners, fat oxidation rose roughly 18% during a 180-minute run at 85 percent of threshold. Spragg's professional cyclists showed the laboratory complement: athletes who oxidized less carbohydrate at moderate intensity had less critical power drop after the fatiguing protocol. Substrate flexibility, specifically the ability to hold fat-dominant metabolism at meaningful absolute workloads, is a durability determinant.

Reading V̇E alongside HR is what lets you separate cardiovascular drift from metabolic demand. If HR rises while V̇E stays flat at the same external workload, you are looking at cardiovascular drift, not deeper metabolic strain. If V̇E and HR both rise, the metabolic cost of the work has genuinely gone up, which is the signature of substrate trouble or threshold erosion.

3. The Neuromuscular System: Breathing Rate and Central Drive

Breathing rate (the frequency of breaths per minute, BR) and tidal volume (the size of each breath, VT) together compose minute ventilation: V̇E = BR × VT. The two components are controlled by separate physiological systems.

Nicolò 2018 established the mechanistic split. Tidal volume tracks metabolic CO₂ production. Breathing rate, in contrast, is regulated by central command and perceived exertion (RPE). The mechanistic relationship is asymmetric: tidal volume continuously adjusts to breathing rate to keep V̇E matched to V̇CO₂, while breathing rate is largely independent of tidal volume.

This is why the rapid-shallow breathing pattern emerges under fatigue. Stevenson 2024 showed that after 2 hours at 90 percent VT1, BR at VT1 increased about 16 percent and tidal volume decreased about 16 percent, while V̇E at VT1 was unchanged. Jaén-Carrillo replicated the same pattern in trail runners: V̇E stable, BR up about 17 percent, tidal volume down about 8 percent during the 180-minute steady-state run, and the same pattern across the uphill time trials.

The implication for durability is direct: as the same external workload demands progressively more central drive to sustain, breathing rate rises. BR drift at a matched workload is therefore a field-measurable proxy for accumulating effort and rising RPE, which tracks neuromuscular fatigue independently of metabolic demand. 

BR is also the "sticky" signal. Heart rate can recover during a recovery interval. Tidal volume can rebound when work eases. Breathing rate, once elevated by neuromuscular fatigue, does not fully reverse within a session. That is what makes it useful as a within-session fatigue read.

How Durability Is Measured In A Lab

The lab gold standard is straightforward: measure a threshold power or speed in a fresh incremental test, perform a standardized prolonged bout (typically 2 to 3 hours of cycling or running), then re-measure the threshold. The shift, in absolute watts or as a percentage of fresh, is the durability number.

This is rigorous and unworkable as a routine training tool. Athletes are not going to do twice-monthly two-hour fatiguing bouts in a metabolic lab.

How Tymewear Measures Durability

Tymewear's VitalPro chest strap captures all three of the durability-relevant signals at once: heart rate, minute ventilation, and breathing rate, second by second, in the field. That is the hardware story. The reason the signals are useful is that each one isolates a different system.

The signals have been validated against laboratory-grade equipment. In a paired session against the Cosmed K5 metabolic cart, Tymewear's V̇E correlates with Cosmed at r = 0.97 on 5-second rolling averages, breathing rate matches within 1.2 br/min on average.

Reading the three signals together is what unlocks the durable-state picture:

• HR drift at a matched workload flags accumulating cardiovascular stress (plasma volume loss, rising core temperature) and correlates strongly with overall durability erosion.
• V̇E at a matched workload is the metabolic ground truth. If V̇E stays flat while HR rises, the metabolic demand of the work has not actually changed, and the HR rise is mostly cardiovascular drift. If V̇E rises with HR, the metabolic cost of the same work has gone up, often because of a substrate shift toward fat or a genuine threshold downshift.
• BR drift at a matched workload flags neuromuscular fatigue and rising central drive. BR is sticky within a session and does not fully reverse, which is what you want from a fatigue signal.

The decomposition matters because different fatigue signatures call for different responses. A session where HR drifts but V̇E and BR stay flat looks like cardiovascular drift on a hot day, not a real durability event. A session where V̇E and BR drift together at a matched workload while HR is moderate is a metabolic-substrate fatigue signature, and points training toward fat-oxidation and aerobic-base work. A session where BR drifts hard while V̇E stays flat is a neuromuscular fatigue signature, and points toward recovery, not more volume.

For threshold-based pacing in particular, watching V̇E behaviour at the workload that previously sat at VT1 is the cleanest way to see whether the threshold has actually moved. If V̇E starts to rise at a power or pace that used to sit comfortably below VT1, the threshold has very likely drifted downward, and the athlete should pull back regardless of what HR or pace says.

How to Train Durability

Research on training interventions for durability is younger than research on durability itself, and most of what exists is hypothesis-driven. Three principles are well supported.

Specificity. To adapt the late-session physiology, you have to expose the athlete to late-session physiology. Some long-duration sessions are necessary because there is no shortcut to the back end of a long race. This is the first message coaches and physiologists tend to converge on.

Pre-loading for time-limited athletes. When a four-hour session is not feasible weekly, creating fatigue early in a shorter session, then putting quality work at the back end, mimics the durable-state demands of a long race. There is a "minimum viable fatigue" idea here: enough preload to actually stress the durable-state system, but not so much that you compromise the quality of the back-end work. Coaches debate the exact placement (fresh intervals first vs. pre-fatigued intervals at the end), and both approaches probably elicit different adaptations.

Substrate flexibility. Spragg's data suggests that athletes who burn less carbohydrate at moderate submaximal workloads are more durable. Jaén-Carrillo's data shows that fat oxidation rises during prolonged work even with carbohydrate fuelling, and that the shift is protective rather than a sign of energetic trouble. Both findings support training that develops fat oxidation capacity at meaningful workloads. That is part of what aerobic-base work is doing:see our complete guide to Zone 2 training. 

There are open questions. Standard training stress metrics (TSS, TRIMP) treat each minute of exercise as equivalent, which they are not. The last hour of a four-hour run is physiologically far harder than the first, and the rate of degradation is individually variable, so a uniform correction factor across athletes is not yet possible. This is one reason within-session signal monitoring (HR, V̇E, BR) is more useful than load summaries alone for durability.

 

Where to Go Next

Durability is broad enough that this page is the entry point, not the full library. From here:

• Cardiac drift in particular, including how to separate benign drift from deleterious drift, is covered in Cardiac Drift Explained: Causes and What It Tells You.
• The fat oxidation side of the story is covered in Fat Oxidation and Durability: Why Substrate Flexibility Matters.
• The pacing application, particularly for ultra events where durability is the dominant variable, is covered in Pacing for Ultra Events: How Durability Translates to Race Day.

For the broader threshold framework that durability sits inside, see the cornerstone on VT1, VT2, and VO₂max. For the marathon-specific application of threshold-based pacing, including the post-race analysis of a runner who used these methods to qualify for Boston, see our intermediate marathon training plan.

To measure your own durability in the field, with HR, V̇E, and breathing rate captured second by second, see the Tymewear VitalPro chest strap.

References

1. Maunder E, Seiler S, Mildenhall MJ, Kilding AE, Plews DJ. The importance of "durability" in the physiological profiling of endurance athletes. Sports Medicine (2021).
2. Rothschild JA, Gallo G, Hamilton K, Stevenson JD, Dudley-Rode H, Charoensap T, Plews DJ, Kilding AE, Maunder E. Durability of the moderate-to-heavy intensity transition can be predicted using readily available markers of physiological decoupling. European Journal of Applied Physiology 125:2911 to 2920 (2025).
3. Spragg J, Leo P, Swart J. The relationship between physiological characteristics and durability in male professional cyclists. Medicine & Science in Sports & Exercise 55(1):133 to 140 (2023).
4. Jaén-Carrillo D, Bruce CD, Gobbo C, Howe C, Lawley JS. Durability in trail running: coupled physiological and biomechanical responses to prolonged submaximal and repeated uphill time trials in trained trail runners. International Journal of Sports Physiology and Performance (2026, in press).
5. Stevenson JD, et al. Ventilatory pattern shift at VT1 during prolonged cycling (2024).
6. Nicolò A, Marcora SM, Sacchetti M. Differential control of respiratory frequency and tidal volume during exercise. European Journal of Applied Physiology (2018).
7. Nicolò A, Marcora SM, Bazzucchi I, Sacchetti M. Differential control of respiratory frequency and tidal volume during high-intensity interval training. Experimental Physiology (2015).
8. Meyer T, Gabriel HHW, Kindermann W. Is determination of exercise intensities as percentages of VO₂max or HRmax adequate? Medicine & Science in Sports & Exercise (1999).
9. Smyth B, et al. Decoupling of internal and external load in marathon runners (2022).
10. Coop J, Maunder E. Koopcast Episode 160: Durability in Ultrarunning.
11. Tymewear vs Cosmed K5 paired-session validation analysis (internal, 2026).