OBJECTIVE--The aim was to examine the relationship between social network and activities and causes of death. DESIGN--The study was a prospective cohort study of middle aged men examined in 1973 and followed for 12 years. SETTING--Gothenburg, Sweden. SUBJECTS--The subjects were 769 60 year old and 220 50 year old men who had participated in a health examination. MEASUREMENTS AND MAIN RESULTS--Main outcome measures were mortality from cardiovascular diseases, cancer and other causes. In multivariate analyses cardiovascular mortality was related to baseline blood pressure (p less than 0.001), smoking habits (p = 0.002), myocardial infarction or stroke (p less than 0.001), and a low level of social activities (p = 0.04). Cancer mortality was related to age (p = 0.003) and smoking habits (p = 0.001). Other causes of death were related to poor perceived health (p = 0.02) and a low level of home activities (p = 0.004). In univariate analyses the above risk factors were strongly related to all three causes of death. In addition cardiovascular mortality was related to age, a low level of home and outside home activities, few persons in the household, and poor perceived health. Cancer mortality was related to poor perceived health, few persons in the household, and a low level of social activities. Other causes of death were related to living alone, poor perceived health, and a low level of social and outside home activities. CONCLUSIONS--Well known risk factors for premature mortality like smoking, hypertension, and major cardiovascular disease are verified. Middle aged men with a good "social network" (here measured as a high level of social, home, and outside home activities) may be partly protected against non-cancer mortality.