The association between circulating cholesterol level and ten-year cancer mortality was investigated in 61,567 men aged 40 to 69 years from 11 population studies in eight countries. Those dying of cancer within one year of cholesterol determination had mean cholesterol levels 24 to 35 mg/dL lower on the average than the rest of the men. For years 2 through 5 and 6 through 10, the inverse association diminished markedly, with differences in mean cholesterol levels of only 4 to 5 mg/dL and 2 mg/dL, respectively. Lung cancer mortality was inversely associated with cholesterol level only in the first year, and no significant association with colon cancer. These findings are consistent with the hypothesis that lower cholesterol levels in cancer decedents are due to the effect of undetected disease on cholesterol level.