Background and aimThe association of serum cholesterol levels with the occurrence of coronary heart disease (CHD) mortality during a follow-up of 50 years was rarely investigated previously. Thus, we took advantage of results at hand in 10 pooled cohorts of men aged 40–59 years from the Seven Countries Study (9063 individuals and 2057 CHD fatal events) and we assessed this. Methods and resultsCox proportional hazards models were run with CHD fatal events (as dependent variable) and cholesterol levels (as independent variables) at years 0, 10, and 25 (in 5 cohorts). Cumulative events during subsequent decades (cumulative approach: CA) and separately in each subsequent decade (partitioned approach: PA) were analyzed.The ecological correlation of average baseline serum cholesterol levels with CHD mortality was very high (R = 0.97). Serum cholesterol and CHD mortality for 50 years were associated at the individual level, and the association estimated by the Cox's coefficients (and related hazards ratios) was initially strong in both CA and PA, but slightly declined during later decades. Hazards ratios (for a difference of 40 mg/dl) ranged from 1.39 to 1.20 for CA and from 1.39 to 0.80 for PA. Coefficients were larger for CA than for PA and the decline was more evident for the latter. Partitioned coefficient became negative and significant in the last decade (from year 40–50). Coefficients derived from cholesterol levels measured at year 10 of follow-up showed similar trends but their magnitude was smaller. ConclusionThus, the relationship of serum cholesterol levels with CHD mortality remained relatively stable during at least 40 years after a single cholesterol measurement at baseline in middle-aged men.