The association between coffee consumption and serum cholesterol was studied in a cross-sectional epidemiological study in Finland where the annual per capita consumption of coffee (13.0 kg) is the highest in the world. Coffee consumption was assessed by a questionnaire in a representative population sample of 4744 men and 4495 women aged 25 to 64 years. Serum total cholesterol and HDL-cholesterol concentrations were determined in fresh sera by the enzymatic method. Data on a large number of potential confounding variables were also collected. In the age group 25 to 44 years, the level of serum total cholesterol increased linearly with increasing coffee consumption in both sexes, but in people aged 45 to 64 the peak level of serum cholesterol was found in those who consumed 4 to 6 cups of coffee per day. In the analysis of covariance controlling for age, body mass index, intake of fat, sugar, and alcohol, smoking, physical activity, and fasting time, the mean level of serum cholesterol of men was lower (p less than 0.001) in those who drank no coffee (5.9 mmol/l) than in those who drank 1 to 3 cups (6.1 mmol/l) or 4 or more cups (6.2 mmol/l) per day. In women, the corresponding mean serum cholesterol values were 5.8 mmol/l, 6.1 mmol/l, and 6.1 mmol/l (p less than 0.05). Serum HDL-cholesterol levels did not vary significantly with coffee consumption. There was a slight inverse association between tea drinking and serum total cholesterol in men (p less than 0.05) but not in women. Although our results suggest a positive association, the impact of coffee drinking on serum cholesterol seems to be minimal. The results also indicate that the possible mechanisms do not include caffeine.