Training load
Training load by system
Most training load numbers measure how much work you did. They don't say what kind: an hour of hard intervals and a long, easy ride can earn the same score while training quite different parts of you.
Read your training by system, against what the session or block set out to train, and use any single total as one input alongside how sessions felt.
At a glance
| The problem | One load total adds up very different kinds of work |
|---|---|
| The systems | Sprint power, anaerobic capacity, maximal aerobic power, threshold, and durability late in long sessions |
| Two useful numbers | Critical power, the limit of steady work, and W′, the work you can do above it |
| Load by system | Some platforms, such as Xert and Wahoo, split each ride's load by the system it stressed |
| The limits | Single scores hide trade-offs, can't see sleep or stress, and rest on debated models |
| What to do | Start from what the block is for, and check where on your power curve the work sat |
Different efforts, different systems
Different efforts lean on different systems. A sprint of a few seconds relies on energy already stored in the muscle and on how forcefully you can recruit it. An all-out minute draws heavily on anaerobic capacity. Efforts of three to eight minutes sit close to your maximal aerobic power, and anything longer depends mostly on your threshold and, late in a long session, on durability. Each quality responds most to work aimed at it, though a strong aerobic base, built mostly below threshold, supports all of them.
For Olympic and 70.3 racing, threshold and durability do most of the work. Short, hard efforts matter mainly on climbs, at the swim start and when you surge.
Why FTP isn't the whole story covers these qualities and how a power profile shows them. Durability covers how well they hold up after hours of work.
Critical power and W′
Plot your best power for every duration and you get your power-duration curve. Between roughly 2 and 20 minutes, researchers describe it well with two numbers. Very short sprints and rides of several hours sit off this simple model.
- Critical power (CP) is the highest power you can hold without fatigue building steadily. A 2016 review describes it as a fatigue threshold: above it, you will have to slow down at a predictable point, typically within about 20 to 40 minutes at CP itself. CP usually sits close to FTP, often a little above it. You can estimate it without a lab from two all-out efforts of about 3 and 12 minutes, or read it from your platform's power curve.
- W′ (W prime) is a fixed amount of work you can do above critical power. Attacks and surges spend it. Riding below CP refills it, and the easier the recovery, the faster it comes back.
Worked example
A triathlete with a critical power of 250 W and a W′ of 20 kJ rides a 5-minute climb at 300 W. That is 50 W above CP, so W′ drains at 50 joules a second: 300 seconds × 50 = 15 kJ, three quarters of their W′. Held for 6:40, it would run out. Back on the flat at 70.3 race power, around 200 W, it refills only slowly, because W′ comes back faster the further below CP you ride.
In non-drafting racing most of your effort sits below critical power, so W′ matters most for hills, surges and short-course racing.
A 2017 review set out how W′ is spent above critical power and rebuilt below it, so riders with the same CP but different W′ handle surges differently. Runners have the same pair, critical speed and D′, and swimmers use critical swim speed (CSS) the same way.
Critical power also moves within a session. In one study summarised in a 2024 review, CP fell by about 10% on average after two hours of heavy cycling, with individual drops ranging from under 1% to about 32%. How much you keep is a quality of its own; see Durability.
Splitting training load by system
Most training load numbers add everything into one total. A training stress score, or session RPE multiplied by minutes, measures how much work you did. It doesn't say what kind.
Same score, different sessions
Some platforms now split load by the system it stressed. Two current examples, described from their own public material:
- Xert splits each ride's strain into three parts named after the systems it models: threshold power, high-intensity energy (Xert's equivalent of W′) and peak power. Strain is scored against the power you have available at that moment, so the same hour at threshold scores higher when you're already tired.
- Wahoo builds on Four-Dimensional Power (4DP), which profiles a rider on neuromuscular power, anaerobic capacity, maximal aerobic power and FTP, tested over 5 seconds, 1 minute, 5 minutes and 20 minutes. Its Dimensional Training Load, introduced in 2025, splits each ride's load across those dimensions, judged against the rider's own profile and their last 90 days. Runs are scored on a three-dimension profile using grade-adjusted pace.
Splitting load this way shows what a session stressed, measured against a model. Whether the block worked shows up later, in your performances and how you feel. Check how each platform describes its method, remember that even TSS depends on an accurate FTP, and test any score against your own experience. Coach Alan Couzens writes that separating volume from intensity improved his own performance models, and that athletes differ in which of the two they respond to.
The trouble with one score
Scores earn their place. They're simple, consistent and quick to read, and a trend can flag something worth a conversation. Device makers adding sleep, heart rate variability and how-you-feel ratings to their apps is a good direction, because more of the athlete is being measured. The trouble starts when it's all rolled into one number.
One total hides trade-offs. A rising number can come from lots of top-end work while threshold and durability work stood still, or the reverse, and for a triathlete one number across three sports hides even more. It also can't see what the work cost you. The same watts cost more on a hot day, after a poor night or at the end of a heavy week at work, and none of that appears in a power file. See Training isn't your only load.
How you feel is evidence too. A systematic review of 56 studies found that athletes' own ratings of mood, stress and wellbeing responded to changes in training more sensitively and consistently than common objective measures such as blood markers and heart rate.
The models behind the totals are debated too. Fitness and fatigue numbers come from a 1975 model whose individual settings are hard to pin down, and one 2021 comparison found that feeding it different load measures, such as TSS or session RPE, shifted its timing by up to 19.6 days. Ratios built on the same totals, such as the acute:chronic workload ratio, have been criticised too. And a single number invites chasing, so you end up training to the score. See Fitness, Fatigue and Form.
The numbers are still useful as one input among several.
How to use this
Coach Gordo Byrn, who still tracks TSS himself, puts it simply: "Let load be the result, not the goal, of training prescription." In practice:
- Keep your totals, and tag them. TSS and CTL still show how much you're doing. Tag each session with its purpose, such as easy aerobic, threshold or hard intervals, and check your weekly hours per tag against the plan.
- Start from the purpose. Before a block, decide which quality it's meant to build. Afterwards, check whether that part of your power-duration curve moved.
- Look at where the work sat. For a VO2 block (hard intervals of 3 to 8 minutes aimed at your maximal aerobic power), compare your best 3 to 8 minute efforts at the start and end, and count the minutes you spent above critical power each week. For a base block, look at your long-ride power late in the session.
- No split-load tool? Use your power curve. Most platforms show a power-duration curve. Compare it month on month, and check time in zone against what each session was for.
- Test with real efforts. A best effort late in a long ride tells you more about durability than any score.
- Run and swim too. In the pool, a 400 m and 200 m test gives your critical swim speed. On the run, track your best 1 km and 5 km, or a recent race.
- Judge the week, not the session. One session rarely tells you much. Look at the whole week, and the block, before you change course.
- Check against how it felt. If the numbers and your session RPE disagree, find out why before you act. See internal and external load.
Common questions
Why isn't one training load number enough?
Because it adds very different work into one total. One hour of hard intervals and two and a half hours of easy riding can both score about 81 TSS while training different systems. It also can't see sleep, heat or a hard week at work.
What are critical power and W′?
Critical power is the highest power you can hold without fatigue building steadily. W′ is a fixed amount of work you can do above it. Surges spend W′, and riding below critical power refills it, faster when the recovery is easier.
What is Wahoo's Dimensional Training Load?
Wahoo's Dimensional Training Load splits each ride's load across the four dimensions of its 4DP profile (neuromuscular power, anaerobic capacity, maximal aerobic power and FTP), judged against the rider's own profile. Runs are scored on a three-dimension profile.
Is TSS useless, then?
No. TSS is a useful, imprecise measure of how much work you did. It depends on an accurate FTP and says nothing about what kind of work it was, so read it alongside your power curve and how sessions felt.
How do I see load by system without special software?
Compare your power-duration curve month on month, check time in zone against each session's purpose, and do occasional best efforts, including one late in a long ride.
Related
Sources
- Poole DC, Burnley M, Vanhatalo A, Rossiter HB, Jones AM. Critical power: an important fatigue threshold in exercise physiology. Medicine and Science in Sports and Exercise. 2016;48(11):2320–2334.
- Jones AM, Vanhatalo A. The 'critical power' concept: applications to sports performance with a focus on intermittent high-intensity exercise. Sports Medicine. 2017;47(Suppl 1):65–78.
- Skiba PF, Chidnok W, Vanhatalo A, Jones AM. Modeling the expenditure and reconstitution of work capacity above critical power. Medicine and Science in Sports and Exercise. 2012;44(8):1526–1532.
- Jones AM. The fourth dimension: physiological resilience as an independent determinant of endurance exercise performance. Journal of Physiology. 2024;602(17):4113–4128.
- Maunder E, Seiler S, Mildenhall MJ, Kilding AE, Plews DJ. The importance of 'durability' in the physiological profiling of endurance athletes. Sports Medicine. 2021;51(8):1619–1628.
- Banister EW, Calvert TW, Savage MV, Bach T. A systems model of training for athletic performance. Australian Journal of Sports Medicine. 1975;7(3):57–61.
- Hellard P, Avalos M, Lacoste L, et al. Assessing the limitations of the Banister model in monitoring training. Journal of Sports Sciences. 2006;24(5):509–520.
- Vermeire K, et al. The influence of different training load quantification methods on the fitness-fatigue model. International Journal of Sports Physiology and Performance. 2021;16(9):1261–1269.
- Saw AE, Main LC, Gastin PB. Monitoring the athlete training response: subjective self-reported measures trump commonly used objective measures: a systematic review. British Journal of Sports Medicine. 2016;50(5):281–291.
- Impellizzeri FM, Tenan MS, Kempton T, Novak A, Coutts AJ. Acute:chronic workload ratio: conceptual issues and fundamental pitfalls. International Journal of Sports Physiology and Performance. 2020;15(6):907–913.
Platforms (their own material)
- Xert. The Xert Strain Score (XSS). Xert support.
- Wahoo. 4DP (Four-Dimensional Power) Cycling Metrics. Wahoo support.
- Wahoo Sports Science. What is Wahoo's Dimensional Training Load and why it's different from TSS. Wahoo blog, 6 November 2025.
- Levitch C. FTP is dead, The Sufferfest claims, and launches 4DP training. BikeRadar, 16 October 2017.
Coaching practice
- Couzens A. Beyond Training Load: Are you a volume or intensity responder?. 21 August 2018.
- Byrn G. How I Track TSS. Substack, 15 August 2024.