The relationship between weight change over a 5-year period and subsequent mortality during a further 4-year follow-up was examined in a prospective study of 7735 British middle-aged men. Over half of the men remained stable (less than 4 per cent change in body weight), 31 per cent gained weight and 14 per cent lost weight over 5 years. There were 357 deaths from all causes and men with stable weight had the lowest mortality rates. Considerable weight gain (greater than 15 per cent gain in body weight) was associated with an increased risk of cardiovascular (CVD) mortality even after adjustment for initial age, body mass index, blood cholesterol, systolic blood pressure and smoking status. Loss of weight was significantly associated with increased mortality largely due to cancer and other non-cardiovascular causes. The association between weight loss and cancer was more marked in non-obese men and emphasizes that weight loss is a potentially serious symptom. Weight loss to non-obese status was associated with a halving of cardiovascular mortality. The benefit was restricted to hypertensive obese men in whom the mortality reduction was considerable and significant. Considerable weight gain in later adult life, even over a short period of follow-up, is not a benign process, it is harmful to health.