Water weight vs fat: which did you just lose?

1. The scale reports mass, and mass is not tissue

A bathroom scale measures the downward force of everything currently inside the body, including the water in the bladder, the food in the gut, the glycogen in the muscle, the water attached to that glycogen, and the plasma volume that a hot session or a salty meal just moved. Fat is one line item among many, and on a twenty-four hour timescale it is the slowest-moving one by an enormous margin.

This is not a subtlety. It is the single largest source of misread data in fight camps, and it produces two opposite failures. An athlete who drops 1.5 kg after a hard Saturday concludes the diet is working and cuts food further, compounding a deficit they did not need. An athlete whose weight sits flat for eight days after a return to lifting concludes the diet stopped working and adds a session, into a week where recovery was already the limiting factor.

Both athletes have the same problem: they are treating a mass reading as a composition reading. The rest of this article is about closing that gap with what is actually available at the side of a gym — a scale, a diary, and arithmetic.

2. Where the 3,500-calorie rule came from, chased to origin

The figure behind every "a pound of fat is 3,500 calories" statement in the sport comes from one paper: Max Wishnofsky, "Caloric equivalents of gained or lost weight," American Journal of Clinical Nutrition, 1958, volume 6, pages 542–546. PMID 13594881.

Two things about that paper are worth knowing, and almost nobody carries them forward.

The first is that it was an arithmetic argument about the energy density of adipose tissue, not an experiment. It reasons from the composition of adipose tissue to a caloric equivalent and arrives at a round number. There is no trial, no intervention arm and no measured weight loss in it. That does not make it wrong — the energy density of fat is a physical property and the estimate has held up reasonably well — but it does mean the rule has always been a modelling assumption wearing the clothes of an experimental result. We have deliberately not reproduced the step-by-step derivation here, because the version of it circulating online is a tertiary restatement rather than the paper's own text, and this site does not print a chain of arithmetic it has not read in the original.

The second is the condition. The rule was derived for a subject with adequate dietary protein who was in glycogen and nitrogen equilibrium — meaning glycogen stores, and therefore the water attached to them, were assumed not to be moving, and lean tissue was assumed neither to be accruing nor breaking down.

Read that condition against a fight camp. Carbohydrate intake is being manipulated deliberately. Glycogen is being depleted in training and partially restored between sessions. Protein intake is being pushed up specifically to defend lean mass under a deficit, which is a nitrogen flux by definition. Fluid intake changes week to week. The one state the rule was built for is the one state a fighter in camp is never in.

So the correct use of 7,700 kcal/kg is narrow and it is still extremely useful: it tells you the most fat that a given energy deficit could possibly account for. It does not tell you what the scale will do.

3. The correction — Hall, and why leanness inverts the intuition

Kevin Hall's 2008 paper in the International Journal of Obesity is the standard correction (PMID 17848938, DOI 10.1038/sj.ijo.0803720). It states the rule of thumb in its usual form — a cumulative deficit of 3,500 kcal per pound of body weight lost, or equivalently 32.2 MJ/kg — and then asks whether the required deficit is actually constant across people.

It is not. Using a modification of the Forbes equation, which makes the composition of the tissue lost depend on how much fat the person started with, Hall finds that a larger cumulative deficit is required per unit of weight lost in people with greater initial fat mass. The classic rule of thumb roughly matches people carrying more than about 30 kg of fat, and overestimates the required deficit for leaner people.

The inversion this produces is the part fighters do not expect. A lean athlete loses a larger share of lean tissue per unit of deficit than a fatter person does, and lean tissue has a much lower energy density than fat and carries water with it. So the lean athlete's scale moves more per calorie of deficit than the 7,700 figure predicts, while the actual fat coming off is less.

For a fighter who is already close to their competitive body composition, this means the scale flatters the diet. A kilogram off the scale in a week at a modest deficit is not a kilogram of fat; a meaningful share of it is lean tissue and its water, which is precisely the tissue the camp is trying to keep. That is the mechanical argument for a protein intake that is defended rather than squeezed during a descent.

Hall's later work with colleagues in the Lancet in 2011 (PMID 21872751) adds the other half of the correction: the body-weight response to a change in intake is slow. Their replacement rule of thumb is that every 100 kJ/day change in intake produces about 1 kg of eventual weight change — and "eventual" is doing real work, because the half-time is approximately one year, with about 50% of the change arriving within a year and about 95% within three. The 3,500 rule assumes the effect of a deficit arrives immediately and never decays. Extrapolate it linearly across months and it over-predicts loss in a measurable direction, which is also the direction a 2014 analysis in the Journal of the Academy of Nutrition and Dietetics reports. We could not read that paper's full text — see the verification section — so we are citing only its direction and none of its figures.

For an eight-to-ten week camp, the half-time caveat matters less than it does for a year of dieting, but it is the reason a camp's early weeks tend to over-deliver on the scale relative to the arithmetic and the later weeks under-deliver. That is not the diet failing. That is the water and glycogen compartment finishing its one-time adjustment, and then the slower process being the only thing left.

4. The ceiling this produces

Here is the arithmetic that does most of the work in this article. Every row assumes 100% of the deficit is met from adipose tissue at 7,700 kcal/kg, which is a physical impossibility in practice — some of the loss is always lean tissue and its water. That is why these are ceilings and not estimates.

Two conditions attach to that table and neither is optional. Fat loss will be lower than these rows, because some of the deficit is always met from lean tissue. And scale loss can be higher than these rows, because the lean tissue that goes brings water with it. The table brackets fat; it does not bracket the scale.

The third condition is the quiet one: the deficit itself is an estimate. No field method measures a fighter's real energy expenditure. Prediction equations carry meaningful error, training expenditure varies enormously between a technical round and a hard one, and food logging in athletes has its own well-documented under-reporting problem — which is a large enough subject to have its own article on this site. An athlete confident they are in a 1,000 kcal deficit may be in 600 or in 1,400.

What is reassuring is that this arithmetic and the sports-nutrition guidance land in the same place. The 2025 ISSN position stand on nutrition and weight-cut strategies for mixed martial arts and other combat sports reports the common recommendation that combat athletes aim for 0.5–1 kg of body mass per week during camp. That sits precisely across the 500 and 1,000 kcal/day rows. The guideline and the physics agree, which is the main reason to trust either.

5. Therefore: anything much above that ceiling is not fat

The strongest version of this article's claim, stated carefully: a single day's scale change of more than roughly 0.1 to 0.2 kg cannot be mostly fat, because no plausible one-day deficit reaches the energy that would require.

Work it in the other direction and the absurdity is clearer. A 1 kg overnight drop would need a deficit of around 7,700 kcal in one day. A 7.25 kg drop over a week would need roughly 55,800 kcal, or about 8,000 kcal per day on top of everything the athlete ate. No human athlete in any sport produces that, and no fight camp comes close.

That second number is not hypothetical. In a 2025 study in Frontiers in Nutrition, 31 professional MMA athletes — 28 men and 3 women, mean age 28 — were monitored on a supervised protocol and lost a mean of 7.25 kg, 10.56% of body mass, across the seven days before the weigh-in. They then regained a mean of 7.5 kg, 11.25% of weigh-in body mass, within 30 hours of the official weigh-in.

Two honest notes on that study. It measured total body mass only; it reports no fat-to-lean split. And it has no control group. It cannot tell you what tissue left — but the energy arithmetic can, and the answer is that essentially none of 7.25 kg in seven days is available to be fat. What it was is water, glycogen and its associated water, and gut content. The proof is that essentially all of it came back in thirty hours.

That "came back" observation generalises. A systematic review and meta-analysis published in Frontiers in Nutrition in 2026 pooled 40 studies of post-weigh-in recovery in combat sports. Its two headline regain pools are not the same pool and should not be quoted as one: the relative regain between weigh-in and competition pooled at 6.23% of body mass across 11 studies and 1,504 athletes, and the absolute regain pooled at 3.71 kg across 12 studies and 1,451 athletes. Separately, in the five cohorts where the same athletes' losses and regains could be matched — 179 athletes in total — the median athlete came back at 106.5% of what they had taken off, above where they started, with a range across those five cohorts of 63.2% to 154.1%.

Three cautions belong with those numbers and all three are the review's own. Heterogeneity across the pooled studies was near-total — I² of 99.2% on the percentage pool and 98.9% on the kilogram pool — which means each pooled mean is a description of a very scattered literature rather than a reliable central estimate, and the 63.2–154.1% range is telling you the same thing. Second, the 106.5% median rests on those five matched cohorts and their 179 athletes, not on all 40 studies and not on the 1,504 of the percentage pool. Third, and most important for an athlete: the authors state plainly that "athletes may regain substantial body mass while remaining hypohydrated or incompletely recovered in neuromuscular, repeated-effort, or glycogen-related outcomes." Urine specific gravity at the official weigh-in pooled at 1.029 — from three studies and 58 athletes only — against the usual euhydration threshold of 1.020. The field arrives at the scale dehydrated, and getting the mass back is not the same event as getting the fluid back into the compartments it came from.

For the diagnostic question this article is answering, the point is narrow and clean. Mass that returns within a day or two of eating and drinking normally was never adipose tissue. Adipose tissue does not reappear overnight, because putting a kilogram of it back on requires roughly as much energy surplus as taking it off required deficit.

6. The three usual suspects, and what each one is worth

Glycogen and the water that travels with it

The ISSN position stand quantifies the whole lever in one sentence: water-bound glycogen stores can be depleted through exercise and carbohydrate restriction, facilitating a 1% to 2% loss in body mass. For an 80 kg fighter that is 0.8 to 1.6 kg, and none of it is fat. It returns when carbohydrate does.

The popular version of the mechanism — "every gram of glycogen holds three grams of water" — is worth correcting, because it is stated as a constant and it is not one.

The 3–4:1 figure originates with Olsson and Saltin's 1970 study in Acta Physiologica Scandinavica (PMID 5475323), which put 19 subjects through prolonged arm and leg exercise, then three days of a fat-and-protein diet followed by four days of a carbohydrate-enriched one, and tracked total body water by tritium dilution alongside muscle biopsies. Thigh glycogen went from 4.5 to 19.9 g per kg of wet muscle. It is a real measurement, but the ratio is a division sum from an estimated glycogen mass in a small sample, not a directly measured binding.

Fernández-Elías and colleagues measured muscle water directly in 2015 (European Journal of Applied Physiology, DOI 10.1007/s00421-015-3175-z) and got an apparent ratio of about 1:3 when subjects stayed dehydrated during recovery, and about 1:17 when the full fluid deficit was replaced. Same protocol, same four hours of recovery, a near-sixfold difference in the ratio driven only by how much the subjects drank. Reviewers have also pointed out that these calculations include all muscle water, not only water bound to glycogen — some of it is unbound water distributed through the muscle, which makes the clean "3 grams per gram" figure an over-simplification of something that has never been directly measured.

Use 1–2% of body mass as the planning figure, because that one comes from a position stand and it is expressed in the unit an athlete actually cares about. Treat the gram ratio as a rough illustration with a wide range attached, or skip it.

Sodium and the extracellular compartment

Sodium moves water, and the size of the effect is individual to a degree that makes any conversion factor dishonest. In one study of 78 subjects with a median BMI of 22.5, changing intake from about 50 to about 200 mmol of sodium per day increased extracellular fluid volume by 1.2 ± 1.8 L (PMID 19282825). Read the standard deviation: it is larger than the mean effect. Some people barely moved; some moved several litres.

Longer metabolic-ward work in nine healthy men across four different sodium levels over 28 days shows weight changes far larger than the mass of the sodium consumed, confirming that the weight is retained water rather than salt (PMID 19173770). Both of these are abstract-level for us and are hedged accordingly.

What is not printable is "one gram of sodium equals X kilograms of water." No primary source supports a fixed ratio, and given a standard deviation larger than the mean, none could.

Gut content and plain throughput

This one needs no citation, only arithmetic: every litre of fluid swallowed is 1.0 kg on the scale until it leaves the body, and every kilogram of food is a kilogram until it does.

The Maurício study is useful here purely as an illustration of scale. Across its supervised protocol, one athlete's fluid intake varied by roughly seven litres per day between the widest-apart days of the study. That is about seven kilograms per day of throughput crossing the scale, in a body whose maximum plausible daily fat change is around 0.13 kg. We are deliberately printing that range and nothing else: the protocol's actual day-by-day volumes are not reproduced anywhere on this site.

That comparison is the reason this article exists. The scale's daily range is dominated by throughput, not tissue, by a factor of roughly fifty. To be completely explicit about what that paragraph is and is not: it is a description of how much mass moves through an athlete in a day, taken from a supervised research protocol. It is not a fluid schedule, it is not a taper, and it is not something to follow. This article diagnoses what already left. It does not tell anyone what should leave next.

7. The fourth suspect nobody expects — post-exercise swelling

The most common reason a fighter concludes the diet has stopped working in week three is that hard or unaccustomed training made them heavier while fat was still leaving.

Eccentric and unfamiliar work produces muscle fibre swelling. Biopsy-level work reports increases on the order of 15% in muscle intracellular water and 15–20% in cell volume after maximal exercise, and — the part that surprises people — fibre size measured at 7 to 8 days post-exercise larger than at 2 to 3 days, by around 24%. The swelling does not peak the next morning and resolve. It can keep rising for the better part of a week. These figures are abstract and secondary level for us (PMCID PMC3626686; Peake JM et al., J Appl Physiol 2017;122(3):559–570, PMID 28035017) and should be read as the direction and rough magnitude rather than as precise values.

Translated to a camp: a heavy sparring week, a return to lifting after a layoff, or a new strength block can hold scale weight flat or push it up for several days while real fat loss continues underneath. The athlete who responds by cutting food further is making a recovery problem worse to solve a measurement artefact. This is also why training load belongs on the same chart as bodyweight — the explanation for a flat week is very often sitting in the load column.

8. The signals an athlete actually has

No fighter has a metabolic ward. Four signals are available at the side of any gym, and together they do the job.

1. Rate. Above roughly 0.2 kg/day sustained, or above roughly 1.4 kg/week, the arithmetic in section 4 says it cannot be mostly fat. Below about 1 kg/week it can be, and that is where the position stand's target sits. This is the strongest single signal because it is the one backed by physics rather than by inference.

2. What happens when carbohydrate and salt come back. Weight that returns within 24 to 48 hours of eating normally again was glycogen, its associated water, and gut content. Adipose tissue does not reappear on that timescale. The post-weigh-in literature is a natural experiment on exactly this point, and the median recovery fraction of 106.5% across five matched cohorts is the cleanest demonstration available.

3. Context. If the drop followed a hot session, a low-carbohydrate day, a low-sodium day, or a long travel day, the first hypothesis is fluid. If a gain followed a hard eccentric session or a return to lifting, the first hypothesis is swelling. Fat is the hypothesis of last resort on any single day, because it is the slowest-moving compartment in the body.

4. The 7–14 day slope, not the morning. Day-to-day noise in body mass is on the order of half a percent of body mass as a one-day standard deviation — about 0.4 kg at 80 kg — and larger in an athlete deliberately manipulating fluid and carbohydrate. A real fat change of 0.13 kg/day is well inside that. A week or two of readings resolves it because the noise is roughly random around the trend while the fat loss is monotonic. The mechanics of doing that properly — what to average, what averaging does and does not buy, and why the noise grows with the interval — are handled in full here and are not re-derived in this article.

Those four together answer the diagnostic question more reliably than any consumer body-composition device does, which brings us to the reason why.

9. Measurement error, and the noise floor

This is the most useful paragraph in the article, so it gets stated plainly.

Field body-composition methods carry an error of roughly three to four percentage points of body fat against a reference method. On an 80 kg athlete, three to four percentage points is 2.4 to 3.2 kg of fat mass. A good week of camp at the recommended rate moves fat by well under a kilogram. The field method's error bar is therefore three to four times larger than the signal it is being asked to detect.

An athlete who "sees" their body fat drop two points in a fortnight is reading noise. Most often it is hydration noise specifically, because bioimpedance is an electrical impedance measurement and the athlete is deliberately changing body water — which is the exact variable impedance responds to.

The published error bars behind that statement:

• Multi-frequency bioimpedance against DXA in a large healthy adult population: mean difference in body fat percentage of about −1.8 with a standard deviation of about 4.1 percentage points overall, and slightly worse in men than in women. The bias is small; the spread is not.
• Bioimpedance in elite athletes, which is worse: bioimpedance spectroscopy underestimated fat mass by about 4.6 percentage points in ice hockey players and about 1.1 points in soccer players against DXA, with agreement at the individual level described as highly variable and directionally unpredictable (PMCID PMC2267441). This is the cohort caveat that matters most here. Athletes are the population in which bioimpedance is least trustworthy, and fighters are athletes who change their body water on purpose.
• Bioimpedance for tracking change: multi-frequency devices track direction similarly to DXA but with much higher variability, and in collegiate American football athletes DXA detected limb fat-free mass increases that bioimpedance did not (PMCID PMC8402408).
• Circumference and tape methods: the US Navy method was validated against hydrostatic weighing at a correlation around 0.90, with a standard error of roughly three to four percentage points of body fat. Military validation work reports circumference-based equations produce the largest errors of the methods compared and underestimate body fat at the start of training in both sexes (PMCID PMC10282178; PMCID PMC9008774).

All of those are abstract or secondary level for us, and the three-to-four point figure should be read as "roughly." We are not printing a headline error figure for skinfolds at all: a defensible standard error of estimate for a trained tester with population-appropriate equations exists in the literature, but we could not pin one to a primary source we actually read, so the only statement this site will make is that skinfold error in practice is of the same order as bioimpedance error. The position stand's own summary is that DXA is considered one of the most accurate methods available, with bioimpedance a practical and cost-effective alternative despite some measurement variance.

One thing the noise floor is not: the scale. We have deliberately not printed an accuracy specification for any consumer scale, because every figure available is a vendor claim and none is independently validated in a published source. What is defensible is the comparison — a household scale's resolution is a fraction of the roughly half a percent of body mass that fluid balance moves on its own overnight. The noise is the body, not the instrument. Buying a better scale does not buy a better answer.

10. A worked scenario: 2 kg overnight

This is a scenario constructed from published figures, not an athlete we have worked with. Fighter Cut has no coached roster and every example on this site is arithmetic.

An 80 kg athlete, eight weeks from a bout, weighs 80.0 kg on Tuesday morning and 78.0 kg on Wednesday morning. Two kilograms in one night. What happened?

The fat ceiling. Two kilograms of adipose tissue is about 15,400 kcal. Tuesday's entire energy expenditure for an 80 kg fighter training twice is somewhere in the region of 3,500 to 4,500 kcal, and they ate something. Even at literal zero intake and a very hard day, the maximum possible fat loss is on the order of 0.5 kg, and realistically — at a plausible 1,000 kcal deficit — it is about 0.13 kg. That 0.13 kg is about 6.5% of the two-kilogram move, roughly one-fifteenth of it. Even at the physically impossible ceiling of 0.5 kg it is a quarter. On the realistic figure, more than 90% of those two kilograms is something else.

What the something else can be, sized. Glycogen plus its associated water is 1–2% of body mass for this athlete, or 0.8 to 1.6 kg, and a hard Tuesday plus a low-carbohydrate Tuesday can move a meaningful fraction of it. Gut content and bladder are straightforwardly a kilogram or more of variation depending on when the athlete last ate and last went to the bathroom. Sweat losses from a hot session are directly a kilogram scale event. Ordinary day-to-day variability, before any of that, is around ±0.4 kg at this body mass.

Add those and two kilograms is not remarkable. It is an ordinary Tuesday with a hard session, a light dinner and a low-carbohydrate day stacked on top of each other.

The test. The athlete eats and drinks normally on Wednesday — normal carbohydrate, normal salt, normal fluid. Thursday morning reads 79.6 kg. That 1.6 kg return in 24 hours is the diagnosis: adipose tissue does not come back overnight, so 1.6 kg of the original 2.0 kg was fluid, glycogen-associated water and gut content. The remaining 0.4 kg is within the noise band and cannot be attributed to anything on the strength of two readings.

What the athlete should conclude. Nothing, from those three mornings. The two-week slope answers the question the overnight reading cannot. If that slope is running at 0.5 to 1 kg per week, the camp is on the position stand's descent rate and nothing needs changing. If it is running at 2 kg per week, the arithmetic says the extra is coming out of water and lean tissue, and the athlete is spending a resource they will want on fight night.

11. Why the answer changes what you do next

The physiological distinction between the two kinds of loss determines when each one can be scheduled, and getting that wrong is the classic failure mode in this sport.

A fat-loss cut is a slow change in tissue. It does not reverse on rehydration, it does not need to be timed to a weigh-in at all, and it is bounded by energy arithmetic — which means it needs weeks. That boundedness is the entire reason the ISSN position stand puts off-camp walk-around weight at 12–15% above the division requirement rather than at 20 or 25%. The stand also sets macronutrient floors during a weight descent — carbohydrate at 3.0–4.0 g/kg, protein at 1.2–2.0 g/kg and fat at 0.5–1.0 g/kg per day — which are there to protect training quality and lean mass while that slow process runs. The week-by-week shape of that descent is a planning problem covered elsewhere on this site.

A water cut is a reversible displacement of fluid. It is timed to a clock, it costs performance while it lasts, and it is undone in hours. The position stand separates the two explicitly: rapid weight loss is the acute 24–48 hour event before the weigh-in, achieved through water and glycogen manipulation; longitudinal fat loss is the 8–10 week camp.

Confusing the two produces one specific, very common failure. An athlete is 4 kg over with six days to go. The arithmetic ceiling says at most about 0.8 kg of that can come off as fat in six days at a sustained 1,000 kcal/day deficit — and even the 1,500 kcal/day row only reaches about 1.2 kg. Therefore somewhere around 3 kg has to come out as water in the final days — a substantially harsher physiological event than the one they thought they were planning, and one that the evidence says they will arrive at the scale dehydrated from, with a weigh-in urine specific gravity of 1.029 pooled across three studies and 58 athletes, against a 1.020 euhydration threshold. What a commission does about that varies, and hydration testing at weigh-ins is its own subject.

The diagnostic in this article is what prevents that arithmetic from arriving as a surprise in fight week. An athlete who knows which tissue has been leaving knows, at week four, whether the plan is on track — because they know the difference between 2 kg of glycogen and water that will come straight back the moment they eat, and 2 kg of fat that is actually gone. That distinction is also the one Fighter Cut's projection is built on: it classifies the rate of loss against the arithmetic rather than counting the raw scale move.

What the diagnostic cannot do is tell an athlete what is safe. No study in this literature established any percentage as safe, none tested a threshold against harm, and the numbers most often quoted as limits are descriptive averages of what professionals were observed to do. That question belongs with a physician, the commission whose rules the bout runs under, and a coach standing in the room.

12. The populations this evidence does not cover

Nearly every cohort behind the numbers in this article is adult, male and professional, with sample sizes between roughly 8 and 31. The MMA cohort study was 28 men and 3 women. The larger MMA datasets in this literature run 91–95% male. The collegiate wrestling material is a men's programme. The bioimpedance athlete validation work is male elite athletes.

Women are not a coefficient applied to that. Total body water as a fraction of body mass differs, the distribution between compartments differs, and menstrual-cycle effects on fluid balance are a real but poorly characterised variable. The honest summary of that literature: body mass and total body water are reported slightly higher in the luteal phase than the follicular phase, with a proposed mechanism involving progesterone-driven activation of the renin–angiotensin–aldosterone system — but the majority of studies reviewed found no significant difference between phases in body weight, body fat percentage, fat-free mass, skinfolds or bioimpedance-measured total body water, and separate work reports that cycle phase does not appear to affect ad libitum rehydration, overall fluid balance or fluid retention at rest or during exercise when fluids are freely available. The literature is small, mixed and mostly in non-athletes, and individual variation exceeds the average phase effect. We are not printing a kilogram figure for it, because there is not one to print. Two or three cycles of a female athlete's own morning data tells her more than any published mean.

Adolescents are a different problem again. Growth is a confound that makes any interpretation of a bodyweight trend over months unreliable in a way it is not for adults, and the combination of a growing athlete and a deliberate energy deficit raises low energy availability questions that sit well outside what a scale can diagnose.

What we could not verify

• Wishnofsky's own derivation. PMID 13594881 resolves and the paper is exactly as cited, but the step-by-step arithmetic chain commonly attributed to it circulates through tertiary sources. We did not read the 1958 text and have not printed the derivation. The "glycogen and nitrogen equilibrium" condition is reported in the secondary literature and is written here as the condition the rule was derived for, not as a direct quotation.
• Thomas et al. 2014 (J Acad Nutr Diet 114(6):857–861, PMID 24857406, PMCID PMC4035446). Full text was unavailable to us on 23 September 2026 — PubMed Central returned a reCAPTCHA challenge and the publisher returned HTTP 403. We cite the direction of its finding only, and print none of its figures. We also decline to print the widely repeated "the rule predicts 50 lb where real loss is 35 lb" pair, which we could not verify against any primary text.
• The Nutrients 2023 review of glycogen and hydration (15(1):155, PMCID PMC9823884), our source for the criticism that glycogen-water ratios include unbound intramuscular water, returned HTTP 403 from the publisher. That criticism is written as "reviewers have pointed out" rather than as a fetched quotation.
• The day-to-day variability figure. Our ~0.5% one-day standard deviation comes from work in Renal Failure 2023;45(2):2273421 (PMID 37955103), which returned a publisher 403 and a PubMed cookie wall on the day. The cohort is also not athletes in camp, where the true variability will be larger. Treat it as a floor. The fuller treatment of that figure, with its real cohort described, is in the trend-reading article linked in section 8.
• The body-composition error figures. The bioimpedance-versus-DXA population comparison, the elite-athlete bioimpedance work, the change-tracking work and the military circumference validations are all abstract or secondary level for us; the publishers' full texts were not obtained. The three-to-four percentage point error figure is written as "roughly" for that reason. We have deliberately not printed a headline skinfold standard error, because we could not pin a defensible figure to a source we read.
• Consumer scale accuracy. Every specification available is vendor-published and none is independently validated in a source we could fetch. We have named no brand and printed no accuracy number.
• The UFC graded figures. Figures often quoted as ceilings — about 6.7% of body mass at 72 hours out, 5.7% at 48 and 4.4% at 24 — began life as a measurement, not a recommendation. They are the average cumulative losses recorded in 616 UFC athletes weighed at those three points before the official weigh-in (Peacock et al., 2022, PMID 36547661), which the 2025 ISSN position stand restates. They describe what professional fighters already do under a promotion's performance staff. Nobody established them as safe, and no study in this literature tested a threshold against harm. They are cumulative and not stages — the 72-hour figure already contains the 24-hour one, which is why they descend — and they must never be summed. The exact decimals we verified inside the position stand rather than inside the source paper, whose own abstract rounds to "nearly 7%."
• The Maurício timing. The published paper gives two irreconcilable timings for the event interval. We print only "within 30 hours of the official weigh-in" and no weigh-in-to-bout interval, because we are not going to choose one for a reader.
• The precision of two figures. The Maurício paper prints 10.56% and 11.25%, and those are the figures used here. Rounded one-decimal versions circulate — including a "11.2%" that mis-rounds 11.25 — and we have not used them.
• The Sun 2026 recovery fraction and its pools. 106.5% (range 63.2–154.1%) comes from five matched cohorts totalling 179 athletes. It is not a result of the full 40-study pool, and it is routinely misattributed to the 1,504 athletes of the relative-regain analysis. The two pooled regain figures are also separate analyses — 6.23% from 11 studies and 1,504 athletes, 3.71 kg from 12 studies and 1,451 athletes — and both carry I² near 99%, so neither is a reliable central estimate.
• What no scale can tell you. No field method available to a fighter can directly measure which tissue left. Everything in this article is inference from energy arithmetic and from what happens next. That is a genuine limitation, not a rhetorical one — the Maurício study itself, with a research budget and a supervised protocol, reported total body mass and no fat-to-lean split.

Sources

Sourced to

1. Caloric equivalents of gained or lost weight — Wishnofsky M, American Journal of Clinical Nutrition, 1958;6(5):542–546, PMID 13594881
2. What is the required energy deficit per unit weight loss? — Hall KD, International Journal of Obesity (London), 2008;32(3):573–576, DOI 10.1038/sj.ijo.0803720, PMID 17848938
3. Hall et al. 2011 web appendix — National Institute of Diabetes and Digestive and Kidney Diseases, supplementary material to Hall et al., Lancet 2011
4. International society of sports nutrition position stand: nutrition and weight cut strategies for mixed martial arts and other combat sports — Ricci AA, Evans C, Stull C, et al., Journal of the International Society of Sports Nutrition, 2025;22(1):2467909, DOI 10.1080/15502783.2025.2467909, PMID 40059405
5. Rapid weight loss and regain in professional mixed martial arts athletes — Maurício CA, Artioli GG, Gonçalves AF, et al., Frontiers in Nutrition, 2025;12:1581698, DOI 10.3389/fnut.2025.1581698
6. Effects of rehydration and nutritional recovery strategies after rapid weight loss on body mass recovery, hydration status, and performance in combat sports athletes: a systematic review and meta-analysis — Sun J, Song H, Sun J, Frontiers in Nutrition, 2026;13:1943678, DOI 10.3389/fnut.2026.1943678
7. Weight loss and competition weight in Ultimate Fighting Championship (UFC) athletes — Peacock CA, French D, Sanders GJ, Ricci A, Stull C, Antonio J, Journal of Functional Morphology and Kinesiology, 2022;7(4):115, DOI 10.3390/jfmk7040115, PMID 36547661
8. Variation in total body water with muscle glycogen changes in man — Olsson KE, Saltin B, Acta Physiologica Scandinavica, 1970;80(1):11–18, PMID 5475323
9. Relationship between muscle water and glycogen recovery after prolonged exercise in the heat in humans — Fernández-Elías VE, Ortega JF, Nelson RK, Mora-Rodriguez R, European Journal of Applied Physiology, 2015;115(9):1919–1926, DOI 10.1007/s00421-015-3175-z
10. Muscle glycogen and hydration: a narrative review — Nutrients, 2023;15(1):155, PMCID PMC9823884 (publisher returned HTTP 403 on 23 September 2026)
11. Rise in extracellular fluid volume during high sodium depends on BMI in healthy men — PubMed record, PMID 19282825 (abstract level)
12. Increasing sodium intake from a previous low or high intake affects water, electrolyte and acid-base balance differently — PubMed record, PMID 19173770 (abstract level)
13. Muscle damage and inflammation during recovery from exercise — Peake JM, Neubauer O, Della Gatta PA, Nosaka K, Journal of Applied Physiology, 2017;122(3):559–570, PMID 28035017
14. Re-evaluation of sarcolemma injury and muscle swelling in human skeletal muscles after eccentric exercise — PubMed Central, PMCID PMC3626686
15. Day-to-day variability of body mass in healthy adults — Renal Failure, 2023;45(2):2273421, DOI 10.1080/0886022X.2023.2273421, PMID 37955103 (publisher 403 on 23 September 2026)
16. Body composition in male elite athletes, comparison of bioelectrical impedance spectroscopy with dual energy X-ray absorptiometry — PubMed Central, PMCID PMC2267441
17. Offseason Body Composition Changes Detected by Dual-Energy X-ray Absorptiometry versus Multifrequency Bioelectrical Impedance Analysis in Collegiate American Football Athletes — PubMed Central, PMCID PMC8402408
18. Body composition changes during 8 weeks of military training are not accurately captured by circumference-based assessments — PubMed Central, PMCID PMC10282178
19. Circumference-Based Predictions of Body Fat Revisited: Preliminary Results From a US Marine Corps Body Composition Survey — PubMed Central, PMCID PMC9008774
20. Changes in body weight and body composition during the menstrual cycle — Kanellakis S, et al., American Journal of Human Biology, 2023;35(9):e23951, DOI 10.1002/ajhb.23951 (publisher 403 on 23 September 2026)
21. Menstrual cycle phase and fluid balance during exercise — Journal of Applied Physiology, 2022, DOI 10.1152/japplphysiol.00580.2022

Read next