Crash diets are good at the thing you expect them to be bad at#
The standard warning is that very-low-calorie diets do not work: you lose fast, you rebound faster, and you end up heavier than you started. Tested head to head, that story does not hold. When researchers randomized 204 adults with obesity to lose 15% of their bodyweight either rapidly over 12 weeks or gradually over 36, the rapid group was far more likely to get there — 76 of 97 (81%) hit the 12.5% threshold versus 51 of 103 (50%) in the gradual group — and after nearly three more years of maintenance the two groups had regained essentially the same fraction of what they lost: 70.5% rapid, 71.2% gradual1.
So the case against crash dieting has to be made somewhere other than the scale, and it can be. The costs are real, they have been measured, and they show up in three places the folklore never mentions: in bone, in an appetite system that is still altered a year later, and in what severe restriction does to eating behavior. Two of the charges usually filed against crash diets — that the weight comes back faster, and that repeated cycles wreck your metabolism — are the ones the evidence does not support. Getting that separation right matters, because the wrong warning is easy to dismiss the moment someone loses 12 kg in ten weeks.
The clean comparison: severe restriction lost more, and cost bone#
The most careful test of severity is the TEMPO Diet Trial, which randomized 101 postmenopausal women with obesity to 12 months of moderate (25–35%) energy restriction on ordinary food, or to 4 months of severe (65–75%) restriction using total meal replacements followed by 8 months of moderate restriction. Both arms were prescribed the same protein intake, 1.0 g per kg of bodyweight per day2.
At 12 months the severe arm had lost 15.3 kg against the moderate arm's 8.4 kg, and 10.2 kg of fat against 5.5 kg. Lean mass fell more in absolute terms (−3.2 vs −2.1 kg) but the losses were proportional to total weight lost, and handgrip strength did not differ between the groups. The finding that separates the arms is skeletal: total hip bone mineral density fell 0.032 g/cm² on severe restriction versus 0.015 g/cm² on moderate (P = 0.002), roughly a doubling.
One disclosure belongs on the record, because this blog applies the same standard to results it likes and results it doesn't: the trial's severe arm ran on a commercial total-meal-replacement product, and its author disclosures include a Nestlé Health Science advisory board seat among several industry relationships. That does not invalidate a randomized trial with DXA endpoints, and the bone finding cuts against the commercial product rather than for it — but a reader should know who sits where before weighing a result that makes meal replacements look effective.
| Outcome at 12 months | Severe (65–75% restriction) | Moderate (25–35%) |
|---|---|---|
| Weight | −15.3 kg | −8.4 kg |
| Fat mass | −10.2 kg | −5.5 kg |
| Whole-body lean mass | −3.2 kg | −2.1 kg |
| Total hip bone mineral density | −0.032 g/cm² | −0.015 g/cm² |
| Handgrip strength | no significant difference between groups |
Read that table honestly and the muscle-wasting story most people expect is not the one the data tell — lean loss tracked the size of the weight loss, which is what any deficit does. The bone column is the one that should change your behavior, particularly if you are postmenopausal, and it is not a cost that shows up in a mirror.
The bill that arrives a year later is appetite#
The more universal cost is hormonal, and it outlasts the diet by a long way. Fifty adults with overweight or obesity completed a 10-week very-low-energy diet and lost a mean of 13.5 kg. Their appetite-regulating hormones shifted exactly as you would predict during the diet — leptin, peptide YY, cholecystokinin, insulin and amylin down, ghrelin up. The finding is what happened next: at 62 weeks, a full year after the weight loss, ghrelin was still significantly elevated relative to baseline, several other mediators were still displaced, and subjective hunger was still significantly higher than before the diet began3.
A year of increased hunger is not a willpower problem and it is not something a person can out-discipline indefinitely. It is the reason both arms of the Purcell trial regained around 71%: the maintenance phase is not a period of coasting, it is a period of eating against a raised appetite signal, which is a specific and different job from losing. The crash diet's real failure mode is that it delivers you to that job faster, with the same appetite headwind and less practice at the daily behaviors that cover it.
The oldest evidence is about behavior, not metabolism#
During World War II, 36 conscientious objectors volunteered for a study of semistarvation and refeeding at the University of Minnesota under Ancel Keys. Most lost over 25% of their bodyweight, and many developed anemia, fatigue, apathy, extreme weakness, irritability, neurological deficits and lower-extremity edema4. The psychological signature — a preoccupation with food that crowded out everything else, and disordered eating on refeeding — is the part that echoed through the following seventy years of eating-disorder research.
The discipline this study demands is the same one owed to any extreme protocol: 36 men, self-selected, on a level of restriction that no commercial diet approaches, in wartime conditions, with no comparison group eating moderately. It establishes that severe restriction can reorganize a healthy person's relationship with food, and how thoroughly. It does not establish that a 1,000-calorie week does the same thing in miniature, and anyone using it that way has skipped the methods. What it justifies is a direction of caution, not a dose-response curve — and it is the reason the behavioral argument against crash dieting is stronger than the metabolic one.
Two charges the evidence does not support#
The first is that repeated loss-and-regain cycles leave you fatter with a slower metabolism. A systematic review of 23 cross-sectional, cohort and interventional studies of weight cycling looked at exactly that. Eighteen studies examined weight or BMI and 13 found no significant association; 15 of 20 found no increase in fat mass; none of 18 found a decrease in lean body mass; and 12 of 14 reported no adverse change in resting metabolic rate5.
The overwhelming majority of evidence suggests that weight-cycling (yo-yo effect) is not associated with any adverse effects in body weight, body composition, and metabolic rate.
Their practical conclusion is worth stating plainly: people struggling with excess weight should not be discouraged from trying again. The second charge is the folk version of metabolic damage — that eating too little makes the body clamp down and hoard fat — which is dismantled in full in the starvation-mode myth. Adaptation to a deficit is real and modest, and it is not the thing standing between you and a result.
So where does that leave the case? Intact, but relocated. A crash diet is not a scam and it is not metabolically ruinous; it is a way of buying rapid loss while spending bone density, arriving at the hardest phase with a raised appetite you will carry for a year, and — the part no trial captures well — skipping the months of practice that produce the habits maintenance runs on. The alternative is not slowness for its own sake. It is a deficit sized so the behaviors survive it, run at a rate whose costs have actually been priced, for long enough that the daily habits become automatic before the diet ends.
FAQ#
Is weight lost quickly regained more quickly?#
Not in the trial designed to test it. Adults randomized to lose 15% of bodyweight rapidly over 12 weeks or gradually over 36 regained nearly identical proportions during a subsequent 144-week maintenance phase — 70.5% and 71.2% respectively — and the rapid group was substantially more likely to reach the target in the first place, 81% versus 50%1. The regain problem belongs to the maintenance phase, not to the speed of the loss.
Does yo-yo dieting wreck your metabolism?#
The evidence says no. A systematic review of 23 studies of weight cycling found no decrease in lean body mass in any of the 18 studies examining it, no adverse change in resting metabolic rate in 12 of 14, and no significant association with body weight or BMI in 13 of 185. The reviewers concluded that people with overweight or obesity should not be discouraged from repeated attempts to lose weight.
How long does appetite stay elevated after a very-low-calorie diet?#
At least a year, on the best available measurement. Fifty people who lost a mean of 13.5 kg on a 10-week very-low-energy diet still had significantly elevated ghrelin and significantly higher subjective hunger 62 weeks after starting, along with persistent displacement of several other appetite mediators3. Plan for maintenance to be effortful for that long rather than treating renewed hunger as a lapse in discipline.
Sources#
- Purcell K, Sumithran P, Prendergast LA, Bouniu CJ, Delbridge E, Proietto J. The effect of rate of weight loss on long-term weight management: a randomised controlled trial. Lancet Diabetes Endocrinol. 2014;2(12):954-62.
- Seimon RV, Wild-Taylor AL, Keating SE, et al. Effect of Weight Loss via Severe vs Moderate Energy Restriction on Lean Mass and Body Composition Among Postmenopausal Women With Obesity: The TEMPO Diet Randomized Clinical Trial. JAMA Netw Open. 2019;2(10):e1913733.
- Sumithran P, Prendergast LA, Delbridge E, Purcell K, Shulkes A, Kriketos A, Proietto J. Long-term persistence of hormonal adaptations to weight loss. N Engl J Med. 2011;365(17):1597-604.
- Kalm LM, Semba RD. They starved so that others be better fed: remembering Ancel Keys and the Minnesota experiment. J Nutr. 2005;135(6):1347-52.
- Sanaya N, Janusaite M, Dalamaga M, Magkos F. The Physiological Effects of Weight-Cycling: A Review of Current Evidence. Curr Obes Rep. 2024;13(1):35-50.
