Why Later Sprint Reps Slow Down: A Fatigue Check

Rep 4 in your sprint set reads 4.62 seconds. Rep 1 was 4.15. That is a real number, and it is tempting to file it under one word — fatigue — end the set, log a hard session, and move on. But a half-second gap on the stopwatch does not tell you what happened inside that sprint, and two athletes losing the exact same 0.47 seconds can be dealing with two completely different problems.

One athlete is producing less force everywhere, from the first stride out of the blocks to the last stride at 30 meters, because the phosphocreatine system and the central nervous system are both running low on the same rep. The other athlete is still producing plenty of force through the first 10 meters, then loses hip position around 20 meters and starts reaching with the lead leg to hold pace. Both show up on the stopwatch as a slower total time. Only one of them is fatigue in the sense that matters for programming, and mistaking the second for the first is how a technical fault gets logged as a normal training response and run into the ground for another six reps.

The number that matters is not how much slower the rep was. It is where inside the sprint the time got lost.

Total sprint time compresses all 30 meters into a single figure, and that is exactly the problem with reading it alone. Two very different physiological events can produce an identical drop in that figure.

Neuromuscular fatigue is a system-wide fall in force output. A single maximal sprint depletes phosphocreatine stores by roughly 60-80%, and with the 20-25 second recovery windows typical of repeat-sprint sets, only about 35-50% of that store is restored before the next rep starts. Girard, Mendez-Villanueva, and Bishop's 2011 review of repeated-sprint fatigue adds the neural side of the same coin: central drive from the motor cortex also declines across repeated maximal efforts, and in protocols with short recovery the review attributes a meaningful, though variable, share of the force loss to that central component rather than to muscle contractility alone. The practical result is a rep that runs uniformly slower, not slower in one specific phase, because the whole system has less to give everywhere at once.

Postural collapse is a mechanical event, not a metabolic one. It shows up as a trunk lean beyond what the athlete used on rep 1, a hip that drops on foot strike, ground contact time that lengthens on one side more than the other, or arm carriage that starts crossing the midline. Schuermans and colleagues (2017) linked exactly this kind of pattern to injury risk in a prospective study of soccer players: altered trunk and pelvis muscle activation timing during maximal sprinting, measured before the season started, distinguished the ten players who went on to strain a hamstring that year from teammates who stayed healthy. The injured subgroup was small, and the measurement was surface EMG in a controlled setting rather than something you can run on a field session — real limits on how far the exact numbers travel. But the underlying point holds up: a technical breakdown at speed is a mechanical failure with its own injury signature, and it is not the same event as simply running out of fuel.

You do not need surface EMG to tell these two apart on a field session. You need three splits per rep instead of one finish time, because fatigue and postural collapse leave different fingerprints across the 0-10m, 10-20m, and 20-30m sections of a standard sprint.

A fatigue-driven rep loses time in roughly the same proportion across all three splits, because the whole system is producing less force out of the start and less force at top speed alike. A postural-collapse rep holds its early splits close to baseline and then falls off sharply in one specific section — almost always the 20-30m segment, where trunk control and hip extension under load matter most — or shows a left-right asymmetry that was not present on rep 1.

Same Total Time, Different Signature

Notice that both signatures in the table land within a tenth of a second of each other on total time. Read the finish line alone and they look indistinguishable. Read the three splits and they are not the same event at all: one is a system running on roughly 93% of its fuel, the other is a hip failing to do its job on the last third of the sprint. The number worth watching is the spread between splits, not the slowest split by itself. A rep where every split sits 6-9% behind rep 1 is a normal fatigue curve. A rep where two splits sit 2-4% behind and the third sits 18% behind is not fatigue — it is one joint or muscle group failing in one part of the movement, and running more reps on that pattern trains the compensation instead of the sprint.

Equipment: dual-beam timing gates or an IMU-based system positioned at 0m, 10m, 20m, and 30m; a flat 35m runway with a 5m run-in so the first gate captures a true standing start rather than reaction time; a stopwatch to standardize the 20-25 second active recovery between reps.

Protocol: 6 x 30m sprints with 20-25 second walk-back recovery, matching the field-based repeat-sprint structure Bishop and colleagues (2001) validated in team-sport athletes. Record all three splits on every rep — not just the finish time — and calculate each split's percent change from its own rep-1 value before moving on to the next rep.

Reading the Numbers

The 15-percentage-point line in that table is not an arbitrary cutoff. Bishop's group reported split-to-split variation of roughly 3-5 percentage points in clean fatigue curves among elite and sub-elite field-sport athletes, so setting the flag at 15 points catches a genuine mechanical outlier while leaving ordinary session noise alone. A left-right contact-time asymmetry past 10% follows the same logic — well outside what shows up between limbs on a rep that is simply tired.

A uniform decrement crossing 10-12% total time is a legitimate stop point for a fatigue-management session. Extend recovery to 45-60 seconds — enough to push phosphocreatine restoration back up toward roughly 65-70% — and one or two more clean reps are usually still on the table; otherwise, end the set and bank the score as a genuine data point, the way the repeat sprint ability test protocol treats a completed fatigue index.

A localized or asymmetric split calls for a different response, regardless of how low the total fatigue index reads: stop the set now. Do not average it away into a session total, and do not treat the next rep as more data — treat it as running the fault into the ground. Check whether the pattern is new or recurring against something like the stride asymmetry wearable data from recent sessions; a pattern that only shows up under late-session load points to a conditioning gap, but one that shows up by rep 2 or 3 most sessions points to a strength or mobility deficit that sprinting through will not fix.

Either way, log which signature triggered the stop, not just the time it happened. Over a training block, a rising rate of postural-collapse stops at earlier rep numbers is one of the more useful early-warning readings available — closer in spirit to the early fatigue warning signs coaches already track through jump height and bar velocity than to a normal RSA decrement, and worth cross-checking against the broader velocity-loss fatigue research before adjusting the week's volume.