In the Twin Cities Prospective Study, executive men aged 45 to 55 and "healthy" at the entry examinations in 1948 were re-examined yearly to 1975. Follow-up through 1983 lost only one man. High density lipoprotein cholesterol (HDL) in the serum was measured in 1955 with a method checked with recent standard methods. Among 217 men, 130 were dead by 1983, 56 from coronary heart disease (CHD) and 27 from neoplasms. Survivors did not differ in mean HDL from the men who died but they had higher values than the men dead from CHD. Men dead from neoplasms had significantly higher HDL than men dead from CHD. Men dying early did not differ in HDL from those dying later but they had higher blood pressures. HDL was unrelated to age at death from all causes but was related to age at death from CHD. HDL was not related to age, total cholesterol, smoking, or respiratory function but was negatively correlated with measures of body fatness. Multiple regression and multiple logistic analyses showed no difference in HDL between survivors and men dead from all causes, but men dead from CHD tended to have lower HDL. The data indicate that longevity is not related to HDL in middle age.