Quantifying Neuromuscular Fatigue from Repeat-Jump Tests

A conditioning coach runs the squad through a repeat-jump test on Thursday, the day after a brutal Wednesday session. Jump height comes back within a centimeter of Monday's baseline. Relief, briefly - until three athletes mention their legs still feel like sandbags and one pulls up tight in the first sprint drill on Friday. The jump test said fine. The hamstring didn't agree. This is the single most common failure mode in repeat-jump fatigue monitoring: picking a metric that doesn't carry the signal you're after, then trusting it because it's the number the software shows first.

Repeat-jump testing - a short set of countermovement jumps spaced across a session, or a continuous run of rebound hops - is popular because it's fast and needs no lab. But fast doesn't mean the raw output tells you what you think it does. Some metrics genuinely track accumulated neuromuscular fatigue. Others move around so much between two rested mornings that a real 8% drop and a coin-flip fluctuation look identical. This guide covers which repeat-jump metrics hold a usable fatigue signal, which are mostly noise dressed up as insight, and how to run and read the test so you're not chasing your own measurement error.

Every jump metric carries two numbers whether anyone reports them or not: how much it typically moves when nothing physiologically meaningful happened (day-to-day noise, usually expressed as a coefficient of variation), and how much it moves when real fatigue is present. A metric only earns a place in your monitoring plan when the second number reliably clears the first. If a metric swings 10% between two rested mornings, a 10% drop after a hard session tells you almost nothing - it's sitting inside the same noise band a good day would produce.

This is the trap that catches most programs building a repeat-jump protocol from scratch: picking the metric that sounds most mechanistically relevant - reactive strength, contraction efficiency - without first checking how noisy it is when nothing is wrong. A physiologically sensitive metric with a wide noise band is often worse for decisions than a duller one that barely moves unless something real changed.

A single repeat-jump trial - five countermovement jumps, or ten continuous rebound hops - produces more than a jump height number once you capture flight time and contact time on every rep. The table below lists the metrics practitioners most commonly pull from that data, what each measures mechanically, its typical day-to-day noise, and how sensitive it has actually shown itself to be to genuine fatigue in the literature.

Notice the pattern: the metric everyone defaults to first, jump height, sits in the bottom row for sensitivity despite having the tightest noise floor. That combination is exactly why so many programs conclude nothing changed when something clearly did.

Gathercole, Sporer, Stellingwerff, and Sleivert (2015, International Journal of Sports Physiology and Performance) put this directly to the test. Working with a group of trained athletes, they induced acute neuromuscular fatigue through a repeated-sprint protocol and compared countermovement jump variables before and after. Jump height showed only a small, inconsistent effect - the kind of change that doesn't reliably separate a fatigued state from normal test-retest variation. Variables built from the shape of the jump rather than its outcome height - flight time relative to contraction time, and eccentric phase duration in particular - showed moderate-to-large effect sizes across the same protocol, picking up a fatigue signal jump height essentially missed.

The mechanistic explanation lines up with how a countermovement jump works: an athlete can partially compensate for reduced neuromuscular capacity by altering strategy - dipping deeper, extending the amortization phase, shifting peak-force timing - and still reach close to the same jump height by trading efficiency for outcome. That compensation shows up in contraction-time and eccentric-duration variables well before the height number moves. The study's own limitation is worth naming: the fatiguing protocol was repeated-sprint work rather than a full match, and the sample was moderate, so the exact effect sizes shouldn't be read as universal constants - but the qualitative pattern, outcome metrics lagging kinetic and temporal ones, recurs consistently enough across the broader CMJ literature to treat as real rather than a one-off.

Cormack, Newton, McGuigan, and Doyle (2008, International Journal of Sports Physiology and Performance) ran the reliability side of this same question, testing single and repeated countermovement jumps to establish typical error and coefficient of variation across a range of CMJ output variables. Jump height came back as one of the more reliable measures in their data, consistent with the tight noise floor seen elsewhere in the CMJ literature. The flight time:contraction time ratio - one of the variables that showed up as fatigue-sensitive in the Gathercole work - came back noisier, with a meaningfully higher coefficient of variation than jump height in the same testing sessions.

Put the two studies side by side and the tradeoff is unavoidable: the metric most sensitive to real fatigue is also one of the noisier metrics day to day, and the metric with the tightest noise floor is one of the least sensitive to real fatigue. Neither paper was designed to resolve that tension directly - Cormack and colleagues tested healthy, non-fatigued athletes across repeated sessions to characterize baseline noise, while Gathercole and colleagues tested a fatiguing intervention without the same multi-week reliability framing - so treat the comparison as two pieces of a puzzle rather than one integrated study. The practical takeaway survives that limitation regardless: a raw percentage drop in flight:contraction ratio means little until you know that metric's own noise floor for your athletes, ideally from your own repeated baseline testing rather than borrowed figures from a different population.

The protocol matters as much as the metric choice. A five-jump or ten-jump set gives you almost nothing useful if rest intervals, cueing, and rep count aren't standardized enough to trust the comparison. Here's a version that holds up across most team and individual settings:

• Standardize a 5-8 minute general warm-up plus 2-3 progressively loaded practice jumps before every testing occasion, identical to your baseline session.
• For a within-session or across-session check: five countermovement jumps, hands on hips, with 15 seconds of rest between reps - close to the Cormack reliability protocol above, and it keeps each rep near-maximal rather than compounding local fatigue into the picture.
• For a within-set mechanical decrement instead: ten continuous maximal rebound jumps with minimal ground contact, capturing flight time and contact time on every rep, not just the first and last.
• Log jump height, flight time, contact time, and eccentric duration for every rep. The signal usually lives in the shape of the decline across reps, not the single best value.
• Calculate a within-set fatigue index where useful: (best rep minus worst rep) divided by best rep, times 100 - more informative on flight:contraction ratio or contact time than on jump height, given where the signal actually sits.
• Establish your own noise floor before drawing conclusions: run the identical protocol on two or three rested, untrained-fatigued occasions and treat that spread as your working coefficient of variation, rather than importing a figure from a different sport or age group.

Our 10/5 repeated jump test protocol and practitioner's guide to assessing fatigue with jump testing walk through variations on this setup in more operational detail.

Once you know a metric's noise floor, the decision rule is simple: don't act on a change smaller than roughly 1.5-2x that metric's own coefficient of variation. A jump height drop of 3% on a 4-5% noise floor is not a finding. A flight:contraction ratio drop of 18% on a 10-11% noise floor is a genuine signal. The table below turns that logic into a rough starting interpretation band per metric - a starting point to replace with your own athlete-specific data, not a fixed clinical cutoff.

Two other reads matter beyond a single threshold. Agreement across metrics beats any one number alone - a jump height drop alongside a lengthening eccentric duration and a falling flight:contraction ratio is a far stronger case than any single metric moving in isolation. And direction matters as much as magnitude: a flight:contraction ratio falling session over session, even while staying inside the single-session noise band each time, is worth flagging before any one point crosses a threshold.

The most frequent error is the opening scenario: reading jump height alone, seeing it hold steady, and clearing an athlete still genuinely fatigued in every metric that would have caught it. The fix isn't dropping jump height - it's no longer treating it as the whole story.

A close second is running a protocol with too few reps to say anything about within-set decrement, then reporting a fatigue index anyway. A three-rep set can't give a stable best-versus-worst comparison; small rep counts can produce a decrement that looks alarming purely by chance. Eight to ten reps, minimum, if within-set decrement is the metric in play.

The third is comparing a flight:contraction ratio or eccentric duration figure against a published benchmark from a different sport, age group, or device rather than the athlete's own baseline spread. Given how much noisier these variables are than jump height, borrowing someone else's noise floor tends to either miss real fatigue or flag phantom fatigue on a normal day. Our neuromuscular fatigue monitoring methods comparison and guide to reading CMJ numbers that drop without matching subjective fatigue both go deeper into building that reference range properly.