Recent studies have suggested that weight loss in middle-aged persons antecedes increased mortality. Therefore, the authors sought to examine the association between changes in body weight and subsequent mortality, according to self-reported dieting status. The authors followed 9,228 men aged 40-65 years in 1963, for whom weight changes between 1963 and 1968 were recorded and extensive clinical, anthropometric, biochemical, and dietary assessments were made. Of these men, 2,471 reported being on a diet when first examined in 1963, and 636 were dieting primarily to lose weight. Mortality follow-up covered an 18-year period (1968-1986). Men who lost 5 kg or more between 1963 and 1968 ("extreme weight losers") exhibited the following age-pooled risks of mortality relative to the stable weight group: for total mortality, 1.36 (95% confidence interval (CI) 1.20-1.55); for all cardiovascular disease mortality, 1.40 (95% CI 1.16-1.69); for all non-cardiovascular disease mortality, 1.33 (95% CI 1.11-1.59); for coronary heart disease mortality, 1.55 (95% CI 1.25-1.93); and for cancer mortality, 0.90 (95% CI 0.65-1.24). After adjustment for differences in coronary heart disease risk factor levels and morbidity between these groups at the end of the weight change period (1968), the excess risks associated with extreme weight loss declined by approximately one third. They declined further if adjustment was made for 1963 (pre-weight-change period) morbidity and risk factor levels. Being on a slimming diet, as reported in 1963, was associated with an approximate doubling of excess mortality in men with extreme weight loss. Weight loss in 1963-1968 coincided with an increased incidence of coronary heart disease and diabetes mellitus and a declining level of serum total cholesterol. This and other studies indicate that both voluntary and involuntary weight loss might be associated with a small increase in the risk of all-cause mortality.