Exercise, Weight Stability, and Longevity: What 6.5 Million Adults Show

Long-term survival tracked best among people who stayed physically active and kept their weight fairly stable. In a Korean cohort of 6,572,984 adults, the group that remained active while keeping weight change below 5% had the lowest mortality over a median 7.4 years.

That is a useful signal, not a prescription to freeze the number on your scale. The study cannot prove that stable weight caused longer life, or that exercise alone would reproduce the observed mortality gap. It does show something worth taking seriously: inactivity paired with substantial weight loss or gain marked the highest-risk groups.

The large cohort found a steep risk gradient

Lee et al.'s 2022 population-based cohort study followed adults across two health screenings and recorded 91,347 deaths. The researchers classified weight stability as a change of less than 5% of body weight and categorized people by whether they remained active, became inactive, became active, or remained inactive.

Using people who remained active and weight-stable as the reference group, every less favorable activity-and-weight pattern had higher adjusted mortality. The largest relative associations occurred among people who remained inactive and lost weight (adjusted hazard ratio 2.30; 95% CI, 2.22–2.38) and those who remained inactive and gained weight (2.17; 95% CI, 2.09–2.25).

Activity still separated risk when weight stayed stable. Compared with people who remained active, adjusted mortality was higher among those who remained inactive (1.46; 95% CI, 1.41–1.50), changed from active to inactive (1.24; 95% CI, 1.19–1.29), or changed from inactive to active (1.15; 95% CI, 1.11–1.20). The confidence intervals for these estimates exclude 1.0, so the associations were statistically significant.

The cohort's lowest-mortality pattern was sustained activity plus weight stability. Its highest-mortality patterns combined inactivity with either weight loss or weight gain.

The authors also reported statistical interaction between activity and weight change: the multiplicative-interaction p value was below 0.001, the relative excess risk due to interaction was 0.38 (95% CI, 0.31–0.46), and the attributable proportion was 0.18 (95% CI, 0.15–0.22). Those results mean the joint pattern related to mortality more than a simple additive reading of the modeled associations would predict. They do not establish a biological synergy, and they do not isolate the causal contribution of exercise from that of weight maintenance.

Why weight loss can look dangerous in mortality studies

Body weight change is not one exposure. A person can lose fat deliberately during a well-designed program, lose muscle during illness, lose fluid during treatment, or lose weight because a disease has already begun to change appetite and metabolism. A scale cannot sort those states for you. It is many things, but it is not a medical detective.

This distinction is especially important for unintentional loss. In an older biracial cohort, Yao et al.'s 2025 Aging Cell study found that a metabolite score linked to unintentional weight loss explained a substantial proportion of the associated risk of mortality and mobility limitation. The study supports reverse causation and illness-related wasting as plausible explanations for some of the mortality seen in weight-loss groups; it does not prove that the metabolites caused those outcomes.

A related 2025 GeroScience cohort analysis by Yao et al. identified circulating metabolomic biomarkers associated with five-year changes in weight and body composition among community-dwelling older adults. It did not test mortality directly. Its value here is narrower: long-term weight change can reflect biologically different states, rather than one clean exposure called "losing weight."

Disease-specific research makes the same interpretive problem harder to ignore. In people receiving hemodialysis, annual dry-weight change predicted later mortality and cardiovascular outcomes, but dialysis adds fluid shifts, protein-energy wasting, and treatment effects. In people repeatedly hospitalized for heart failure, weight change associated with in-hospital mortality, where congestion, acute illness, and cachexia can all alter body weight. These studies help explain why weight change can flag clinical instability; they do not directly replicate the Korean general-population finding.

Intentional weight loss is a different question

The Korean cohort should not be read as evidence that deliberate weight loss harms people with obesity. Randomized and intervention evidence points in a different direction for selected populations and different endpoints.

Bragg et al.'s randomized trial analysis compared exercise plus weight maintenance with exercise plus intentional calorie-restriction weight loss in older adults with obesity. Intentional loss paired with exercise improved intermediate cardiometabolic risk measures. Those are not mortality outcomes, but they directly challenge the simplistic claim that all weight loss is harmful.

Other intervention studies support that distinction. A 12-month exercise-and-diet intervention in older adults with obesity improved the adiponectin-to-leptin ratio in Senkus et al.'s study. A Look AHEAD subgroup analysis examined how body-composition changes after intensive behavioral weight loss related to later physical function. These results concern biomarkers and function, not all-cause mortality, but they show why intentional loss under structured conditions cannot be equated with spontaneous loss in a survival cohort.

Regular exercise also has independent biological credibility. A meta-analysis comparing high-intensity interval exercise with moderate-intensity continuous training found that both approaches improved cardiometabolic risk factors across healthy and clinical populations. That evidence supports exercise as a reasonable target, while still falling short of proving why the Korean cohort's mortality differences occurred.

Stable scale weight can still hide risk

Weight stability is a useful population marker, but it is not a complete health assessment. People can maintain the same body weight while losing lean mass, accumulating visceral fat, or changing fluid status. In type 2 diabetes, Orsi et al. found that waist circumference and other central-adiposity measures predicted mortality independently and may discriminate risk better than body mass index.

Biology can also resist stable weight after deliberate loss. In 80 postmenopausal women with overweight or obesity, Spranger et al. linked a thrifty energy phenotype to greater weight regain after a short dietary program. Regain is not proof of poor adherence; energy adaptation can contribute.

What to do with this signal

Protect activity first. Avoid progressive weight gain, but treat unexplained weight loss as a reason to investigate rather than congratulate yourself. If you pursue intentional weight loss, pair it with exercise and track function, lean mass where feasible, and waist-related risk alongside scale weight.

The evidence supports a restrained conclusion: sustained activity plus relatively stable body weight marked the lowest mortality in this very large observational cohort. The finding is strong enough to guide attention, but not strong enough to prove that changing exercise or body weight alone caused the survival difference.