The association between serum total cholesterol and cancer mortality was investigated in a 28-year follow-up study of 3,091 Dutch civil servants aged 40-65 years and their spouses. After adjustment for age, body mass index, and smoking, an inverse association between serum total cholesterol and all-cancer mortality was observed in men during the first 15 years of follow-up, excluding the first 2 years: The multivariate rate ratio of those in the lowest cholesterol quintile compared with those in the highest quintile was 2.1 (95% confidence interval (Cl) 1.1-4.0); this association was accounted for mainly by gastrointestinal cancers (RR = 4.2, 95% Cl 1.4-12.9). For follow-up years 16-28, the multivariate rate ratio for all-cancer mortality was 0.8 (95% Cl 0.5-1.40. In women, no consistent association between serum total cholesterol and cancer was observed. After the first 5 years of follow-up were excluded, the significant inverse association in men still persisted. These results suggest that in this study population, low serum cholesterol seems not to be attributable to undetected cancer, but might be a risk factor for cancer mortality in men.