The health implications of body weight fluctuation were examined using data from 846 men with a mean age of 56.4 years, enrolled in the Baltimore Longitudinal Study of Aging. Body weight fluctuation was studied in relation to three general categories of health-related outcome: (i) basal metabolic rate, (ii) risk factors for coronary heart disease (CHD), and (iii) rates of CHD, cancer mortality, and all-cause mortality. Weight fluctuation was defined as intraindividual variability in body weight about a time-dependent regression slope. Outcomes (i) and (ii) were described in terms of a subject's rate of change during the same time period. All analyses were adjusted for three covariates: age, level of obesity and linear trend in body weight over time. (i) Body weight variability is positively associated with changes in metabolic rate expressed either per surface area or per lean body mass (i.e. less decrease over time among the most weight-variable men). (ii) Body weight variability is associated with greater decreases in glucose tolerance and with greater increases in the ratio of subscapular to triceps skinfolds, but not with changes in systolic blood pressure, waist-hip ratio, serum cholesterol or triglyceride levels. (iii) Body weight variability is not predictive of subsequent CHD, cancer, or all-cause mortality. These results do not confirm previous findings regarding weight fluctuation and 'hard' endpoints but do raise the possibility that weight cycling might modify eventual cardiovascular risk via adipose redistribution or glucose tolerance impairment.