Fluctuation in body weight is a common phenomenon, due in part to the high prevalence of dieting. In this study we examined the associations between variability in body weight and health end points in subjects participating in the Framingham Heart Study, which involves follow-up examinations every two years after entry. The degree of variability of body weight was expressed as the coefficient of variation of each subject's measured body-mass-index values at the first eight biennial examinations during the study and on their recalled weight at 25 years of age. Using the 32-year follow-up data, we analyzed total mortality, mortality from coronary heart disease, and morbidity due to coronary heart disease and cancer in relation to intraindividual variation in body weight, including only end points that occurred after the 10th biennial examination. We used age-adjusted proportional-hazards regression for the data analysis. Subjects with highly variable body weights had increased total mortality (P = 0.005 for men, P = 0.01 for women), mortality from coronary heart disease (P = 0.009 for men, P = 0.009 for women), and morbidity due to coronary heart disease (P = 0.0009 for men, P = 0.006 for women). Using a multivariate analysis that also controlled for obesity, trends in weight over time, and five indicators of cardiovascular risk, we found that the positive associations between fluctuations in body weight and end points related to mortality and coronary heart disease could not be attributed to these potential confounding factors. The relative risks of these end points in subjects whose weight varied substantially, as compared with those whose weight was relatively stable, ranged from 1.27 to 1.93. Fluctuations in body weight may have negative health consequences, independent of obesity and the trend of body weight over time.