Your reps paid the tax. Your max walked free. And the water level in your bottle had almost nothing to do with it.
Dehydration dropped endurance 8.3 percent and strength 5.5 percent. Vertical jump went up.
The bar didn't feel heavier. Your squat moved, your bench pressed, your overhead felt the same. But the reps cut short, seven where ten should have been routine, and you racked it thinking it was just a bad day.
It probably wasn't random.
A team led by Félix-Antoine Savoie pooled 28 studies, 85 separate tests across 284 athletes, mostly men in their mid-twenties, and measured exactly what mild dehydration costs in the gym. The cost was real. But it was not where anyone expected.
The damage was uneven. Muscle endurance dropped 8.3 percent, roughly one rep vanishing from every ten-rep set. Strength dropped 5.5 percent, barely enough to register on a heavy squat, less than the smallest plate change in most gyms. Your reps went first. Your max barely flinched.
That split changes what dehydration means mid-session. The set that felt like it ended too early probably did. The heavy single that moved fine probably still would have. The cost was real, but it landed on volume, not intensity.
Five outcomes measured. Three fell by different amounts. One was borderline. One went up.
Vertical jump, the one test of raw explosive power that should have gone down with everything else, quietly rose. Not down. Up. Every hydration article you have ever read treated all of this as a single cost. It isn't.
How you got dehydrated mattered nearly three times more than how much water you lost — and the workout itself was the most expensive source.
- The amount of water athletes lost predicted almost none of the performance damage — the dose-response across 28 studies was essentially flat.
- Exercise-induced dehydration hit nearly three times harder than passive dehydration from heat exposure or skipping fluids.
- Trained athletes lost roughly half the performance of untrained participants under the same conditions, though the evidence stopped just short of certainty.
- Anaerobic power — the explosive output in sprints and short bursts — dropped 5.8 percent, the third performance type to take a measurable hit.
- The researchers themselves flagged a concern: participants always knew they were dehydrated, and that knowledge alone may have contributed to the measured cost.
Where the Cliff Should Have Been
If dehydration takes your reps before your max, the obvious next question lands: how bad does it get when you are really dehydrated? Lose three percent of your body weight instead of two, and the damage should multiply. That is the model the entire fitness industry built its hydration rules on.
The model failed.
Savoie's team ran a regression across all 28 studies, plotting the level of dehydration against the size of the performance loss. The amount of water lost predicted one percent of endurance performance and three percent of strength performance. The other 97 to 99 percent of the variation came from somewhere else entirely.
The two-percent threshold, the one the American College of Sports Medicine cemented in their 2007 Position Stand [1], did not show up in the data. Not as a cliff. Not as a curve. Not as anything the analysis could detect.
Even the threshold's strongest defenders reached the same wall. Samuel Cheuvront and Robert Kenefick, two of the most-cited hydration researchers in the field, reviewed the evidence for a dose-response in strength and power and found no clear threshold and no plausible mechanism.
They didn't write that as outsiders. They are co-authors on Savoie's meta-analysis, the same paper this page covers.
The gallon jug on the gym floor, the one with hourly marks and motivational text, is not wrong. Staying hydrated during training still matters. But the formula underneath it, the idea that more water lost means proportionally more damage, explained almost nothing when 28 studies tested it.
What Mattered More Than the Amount
If how much water you lost barely predicted the damage, something else did. The meta-analysis found it: how you got dehydrated mattered 2.76 times more.
When athletes dehydrated through exercise, the thing that actually happens during a training session, performance dropped by an additional 5.4 percent on top of whatever baseline cost dehydration carried. When athletes dehydrated passively, by sitting in heat or simply not drinking, the additional cost was roughly a third of that.
Here is the practical version. If exercise-induced dehydration costs roughly one rep per ten-rep set, the same level of dehydration from a pre-workout sauna costs roughly a third of a rep.
Same water lost. Same number on the scale. Different cost to your training.
The difference was not marginal. The additional impairment from actively sweating through a workout cleared the significance bar comfortably, while passive dehydration methods, including simple fluid restriction, clustered together at a markedly lower cost.
For the lifter checking their water bottle between sets: the dehydration that matters most is the kind already happening to you. Not the sauna before the session. Not the hours since your last glass. The workout itself is the dominant dehydration source, and the biggest performance cost.
Vertical jump quietly improved while every other measure dropped. Dehydration reduced body mass enough that the weight advantage outweighed the minimal loss in force production — for the one test where being lighter actually helps, the physics worked in the athletes' favour.
The Protection That Almost Proved Itself
After the method, the next question is personal. Does being trained change anything?
It does. Or nearly does. Trained athletes lost 1.76 times less performance than untrained participants under the same dehydration conditions. The gap was 3.3 percentage points — meaningful in a training session.
But the confidence fell just short of conventional certainty. The trend was strong, leaning heavily toward trained athletes being partially protected, but the data did not quite clear the threshold the field uses to call something proven.
That near-miss is the honest answer. Not dressed up as significant. Not dismissed as nothing.
What Even the Scientists Weren't Sure About
Savoie and colleagues wrote a line most hydration coverage skips: the nocebo effect cannot be discarded.
In most of the studies they pooled, athletes knew they were dehydrated. That knowledge alone might have cost performance. If you believe dehydration hurts you, part of the measured cost might be the belief doing the work, not the water deficit.
One blinded study offered a partial answer. James and colleagues tested seven cyclists who did not know whether they were dehydrated or fully hydrated, and found performance still dropped about eight percent even with expectation removed [2]. Small sample, extreme heat, but the cost showed up even when the athletes could not tell what condition they were in.
Both pieces together: the cost is real AND partly uncertain. Some of it is physiology. Some of it might be expectation. The data is honest about what it cannot fully separate.
A separate meta-analysis by Deshayes and colleagues, pooling data from 147 athletes, added another layer: the feeling of effort barely increased until dehydration hit about three percent of body weight [3]. Performance was already dropping at lower levels. The tax was being collected before the body registered it.
Then there was the full picture. When every measurable factor went into one model, that model explained 21 percent of what happened. Seventy-nine percent of what determines how dehydration affects your workout is something these 28 studies never measured.
That number is not a failure. It is the finding. Your body is more variable than any hydration formula can account for. The lifter next to you, at the same body weight, losing the same amount of water, may lose a completely different amount of performance. Individual variation dominates every other factor.
When a separate team measured water turnover in 5,604 people across 26 countries, nine variables — activity, body composition, climate, altitude — explained 47 percent of why individuals differ. The best prediction model in the field still missed the majority. Savoie’s 79-percent gap and Yamada’s 53-percent gap point the same direction: your body is more variable than any formula can capture.
The amount of water you lost predicted one percent of the damage. Ninety-nine percent was something else.
Five Outcomes, No Formula
Five performance types, three moderators, and one model that explained a fifth of the picture. The data answered the question behind every marked water bottle: yes, dehydration costs you — but it costs your reps, not your max, and no formula predicts how much.
The next question writes itself. If the amount doesn't matter as much as everyone thought, and the workout itself is the biggest dehydration driver, should you drink on a schedule or just when you are thirsty? A separate meta-analysis set out to answer exactly that.
The next time a set ends two reps short and the weight felt the same, the data from 28 studies suggests a question worth asking: when did you last drink something?
The finding that changes the most is not the size of the cost — it is where the cost landed. Reps absorbed the tax. Max strength barely moved. That split means a dehydration signal does not look like a weaker lift. It looks like a shorter set.
The other shift is about the source. The workout itself — not the hours before it, not the sauna, not the skipped glass — was the single most expensive form of dehydration the meta-analysis measured. The cost was accumulating during the session, not before it.
None of this prescribes how much to drink. What it offers is a calibrated lens: which part of the session is suffering, and which kind of dehydration is doing the damage.
What other research found
What this means for you
The biggest hit in the meta-analysis landed squarely on your kind of training. Muscle endurance dropped 8.3 percent — roughly one rep vanishing from a ten-rep set.
The tax applied equally to upper and lower body. Arms, legs, shoulders — the data showed no significant difference between muscle groups. Every set in the session paid the same rate.
The lifter chasing heavy singles barely noticed. Your session structure — high volume, moderate load, grinding reps — is where dehydration collects its toll.
Your max is relatively protected. Strength dropped 5.5 percent across 39 comparisons — on a 200-kilogram squat, that is about 11 kilograms, less than the smallest plate you would add.
Vertical jump, the closest test to raw explosive power, was completely unaffected. The force that launches you off the ground did not bend under the same dehydration that cut endurance reps short.
But if your programme includes accessory sets at higher reps — and most strength programmes do — those sets are taxed at the higher endurance rate. The max moves fine. The back-off sets might not.
The meta-analysis found a trend the experienced lifter wants to hear: trained athletes lost roughly half the performance that untrained participants did under the same dehydration.
The difference was 3.3 percentage points — meaningful in a gym session. But the evidence stopped just short of the line scientists use to call something proven. The trend was strong. The certainty was not quite there.
What that means in practical terms: years of adaptation appear to offer some armour against dehydration's cost. The less-trained lifter beside you probably loses more from the same water deficit.
The meta-analysis broke dehydration into categories — and yours is the cheap kind. Passive heat exposure cost about 4.1 percent. Exercise-induced dehydration cost 9.4 percent. The difference was statistically significant.
Fluid restriction — just not drinking — landed in the same low-cost range as the sauna, at 4.0 percent. The two passive methods were statistically indistinguishable from each other.
If there is one thing to pay attention to, the data points at the session itself. The dehydration that matters most is the kind accumulating while you train — not the kind from the sauna beforehand.
Before you change anything
The study pooled data from 284 participants, 94 percent of them male, with a mean age of 24. If you are a woman reading this, the findings are drawn almost entirely from young men — the data may or may not apply to you in the same way.
Every participant was healthy. About 45 percent were trained athletes and 55 percent were untrained. The training moderator partially accounts for this split, but nobody studied chronically trained elite athletes separately.
The dehydration protocols were acute and lab-based — athletes were dehydrated before testing, not during a real workout where fluid loss builds gradually. That gap between laboratory and gym floor matters for interpreting the numbers.
One finding is fragile. Anaerobic power dropped 5.8 percent across nine comparisons — but removing a single study reduced the effect to 3.8 percent and pushed the confidence interval across zero. That specific number rests on thin ice.
The endurance category combined different types of tests — isometric holds and repeated contractions grouped together. The heterogeneity was very high, meaning the individual studies varied enormously in their results. The headline number of 8.3 percent is the average across that wide spread.
Every study used acute dehydration in a lab, not the gradual fluid loss that happens during a real training session. The practical translation — one rep per ten-rep set — assumes the lab cost transfers directly to the gym floor.
The headline numbers for endurance and strength are robust. Twenty-eight studies, leave-one-out analyses stable, no significant publication bias detected by statistical testing.
The dose-response non-finding is equally solid: 28 data points and the relationship was flat. The active-versus-passive distinction cleared the significance bar comfortably.
The training moderator is the softest finding. The trend leaned strongly toward trained athletes being protected, but it stopped just short of the significance threshold. Strong enough to notice. Not yet strong enough to call proven.
No publication bias was detected for any of the primary performance variables — endurance, strength, or vertical jump. The pooled estimates are not inflated by selective publication of positive results.
The one question 28 studies did not touch: does the drinking strategy matter? If the amount you lose barely predicts the damage, and the workout itself is the most expensive source — should you drink on a schedule, or just when you are thirsty?
A separate meta-analysis set out to answer exactly that, comparing ad libitum drinking against programmed hydration plans during exercise. The answer was not what either camp expected.
What This Study Found
All findings from this paper, in plain language.
- Dehydration cut muscle endurance by about eight percent, roughly one rep disappearing from every ten-rep set.
- Muscle strength dropped about five and a half percent, a smaller hit than endurance and barely noticeable on a heavy single.
- Explosive power in short bursts dropped about six percent, though this finding depended heavily on one study.
- Total work output in all-out sprints showed a small drop that did not reach significance, leaving the cost uncertain.
- Vertical jump height was not hurt by dehydration and may have slightly improved because losing water weight made athletes lighter.
- How dehydrated athletes were predicted almost nothing about how much performance dropped.
- Dehydrating through exercise hit nearly three times harder than dehydrating passively through heat or skipping fluids.
- Trained athletes lost roughly half the performance of untrained participants, though the evidence did not quite reach significance.
- Every measured factor combined explained only about one-fifth of the variation in how dehydration affected performance.
- The researchers flagged that athletes always knew they were dehydrated, and that awareness alone may have worsened their performance.
- No evidence of publication bias was found — the headline numbers were not inflated by selective reporting of positive results.