Caloric Restriction and Longevity: What 80 Years of Research Shows

The Animal Record and Its Cracks

McCay's 1935 rat experiments established the pattern: restricted animals lived longer and showed delayed onset of age-related disease. Across subsequent decades the effect was reproduced in many species and generally scaled with the degree of restriction up to a point, beyond which malnutrition dominated.

Two later findings complicated the picture. In a large study across dozens of genetically distinct mouse strains, restriction extended lifespan in most but shortened it in some, indicating that genotype determines the direction of response. And in several models the benefit tracked protein or specific amino acid intake more closely than total calories, which reframes what the intervention is actually doing.

The rhesus macaque trials are the instructive case. The Wisconsin study reported a survival benefit; the National Institute on Aging study did not. Reconciliation came from the diets: the NIA control animals were fed a healthier diet and were not overfed, while the Wisconsin controls ate freely from a diet high in sucrose. The Wisconsin benefit was partly the difference between restriction and overconsumption, not between restriction and adequacy.

That distinction runs through the whole literature. Much of what looks like a benefit of restriction is the absence of chronic energy excess, which is a considerably more modest claim and a more achievable one.

CALERIE: The Human Trial

CALERIE II randomised 220 non-obese adults to 25 per cent caloric restriction or ad libitum eating for two years. Achieved restriction averaged around 12 per cent, which is itself a finding: sustained restriction is hard even in a supported trial setting.

Results in the restricted group included reduced body weight and fat mass, lower blood pressure, improved lipid profile, lower fasting insulin and improved insulin sensitivity, reduced high-sensitivity CRP and other inflammatory markers, reduced thyroid hormone T3 and lower resting metabolic rate adjusted for mass, plus improved self-reported mood and sleep quality despite expectations of the opposite.

A later analysis of stored samples reported slowed pace of ageing on DunedinPACE in the restricted group, without a corresponding effect on first- or second-generation methylation clocks. That is one of very few randomised demonstrations of any intervention moving an ageing measure, and it is worth holding at its actual weight: one trial, one measure among several tested, modest effect size.

The costs were also measured. Bone mineral density fell, lean mass fell, and the reduction in T3 and resting metabolic rate is a metabolic adaptation that makes weight regain more likely if restriction stops.

Mechanism: What Restriction Is Signalling

Restriction is not a single mechanism, it is a shift in nutrient-sensing state across several linked pathways.

Reduced mTOR signalling lowers the growth and biosynthesis programme and permits autophagy. Amino acid availability, particularly leucine, is the dominant input here rather than calories as such.

AMPK activation follows from a lower cellular energy charge and shifts the cell toward catabolism and mitochondrial biogenesis. This is the AMPK arm of the same nutrient-sensing system.

Sirtuin activity rises with NAD+ availability, which increases when energy is scarce.

Reduced insulin and IGF-1 signalling is the pathway with the most direct genetic evidence for lifespan effects across species.

Lower inflammatory tone, partly through reduced visceral adiposity, which is an endocrine tissue rather than an inert store.

The convergence is what makes restriction interesting mechanistically and what makes it hard to replicate pharmacologically: several compounds each touch one arm of it, and none reproduces the coordinated shift.

Costs, and Who Should Not Do This

Restriction is inappropriate for anyone underweight, pregnant or breastfeeding, with a history of an eating disorder, in adolescence, in older age where sarcopenia and frailty dominate the risk picture, or with a condition where energy availability matters clinically. That is a wide set of exclusions.

The older-age point deserves emphasis. In later decades, maintaining muscle mass and adequate protein intake predicts function and survival better than restriction does, and the intervention that helps a 40-year-old may harm a 75-year-old.

What the Evidence Actually Supports

Stated as precisely as the data allow.

Well supported. Avoiding chronic energy excess improves metabolic and inflammatory markers and reduces incidence of age-related disease. Most of the human benefit attributed to restriction sits here.

Reasonably supported. Modest sustained restriction in overweight or normal-weight adults improves cardiometabolic markers, and in one trial slowed a pace-of-ageing measure.

Weakly supported. That restriction extends human lifespan. No trial can currently show this, and the animal genotype-dependence makes extrapolation uncertain.

Not supported. That severe restriction is better than moderate, that a supplement reproduces the effect, or that restriction benefits everyone regardless of age and body composition.

The practical translation for most people is unglamorous: eat to maintain a healthy body composition rather than in surplus, protect protein intake and resistance training so that lean mass and bone are not the currency paid, and regard the more aggressive protocols as an experiment with real costs rather than an obvious good.

Restriction Mimetics and Adjacent Approaches

Because sustained restriction is difficult and carries costs, most attention has moved to approaches that engage parts of the same signalling.

Time-restricted eating and intermittent fasting shift nutrient-sensing state periodically without requiring a permanent energy deficit, and in most human trials their metabolic benefits are largely explained by the reduction in intake they produce.

Protein and methionine restriction reproduce parts of the animal effect and conflict directly with the muscle-preservation priority that matters in later life, which is an unresolved tension in the field rather than a settled trade-off.

Pharmacological mimetics, principally rapamycin on the mTOR arm and metformin on the AMPK arm, each engage one pathway. Neither reproduces the coordinated shift, and both carry their own considerations.

The honest summary of 80 years is that restriction identified the nutrient-sensing pathways that govern ageing rate, which is a large scientific contribution, and did not deliver a protocol most people should follow.