OBJECTIVE: To investigate the association of plasma insulin with all-cause, cardiovascular, and noncardiovascular mortality. RESEARCH DESIGN AND METHODS: We studied 22-year mortality data from the Helsinki Policemen Study The study population comprised 970 men, 34-64 years of age, who were free of coronary heart disease, other cardiovascular disease, and diabetes. Area under the insulin response curve (AUC insulin) during an oral glucose tolerance test was used to reflect plasma insulin levels. RESULTS: During the follow-up period, 276 men died: 130 from cardiovascular and 146 from noncardiovascular causes. The hazard ratio (HR) for hyperinsulinemia (highest AUC insulin quintile vs. combined lower quintiles) with regard to all-cause mortality adjusting for age, was 1.94 (95% CI 1.20-3.13) during the first 10 years of the follow-up period and 1.51 (1.15-1.97) during the entire 22 years; adjusting for other risk factors, the HR was 1.88 (1.08-3.30) and 1.37 (1.00-1.87) during 10 and 22 years, respectively The corresponding HRs for cardiovascular mortality during 10 and 22 years were 2.67 (1.35-5.29) and 1.73 (1.19-2.53), respectively, for age-adjusted and 2.30 (1.03-5.12) and 1.39 (0.90-2.15), respectively, for multiple-adjusted HRs. A U-shaped association was observed between insulin and noncardiovascular mortality, multiple-adjusted HRs for lowest and highest versus middle AUC insulin quintiles were 1.85 (1.20-2.86) and 1.43 (0.91-2.24), respectively CONCLUSIONS: Hyperinsulinemia was associated with increased all-cause and cardiovascular mortality in Helsinki policemen independent of other risk factors, although these associations weakened with the lengthening of the follow-up period. The association of insulin with noncardiovascular mortality was U-shaped.