Using data from 140,000 men and 32,000 women in the UK and 30,000 men and 12,000 women in Japan, we performed spectral analyses to test for seasonal variation in serum cholesterol levels. In both countries and both sexes we found a strong seasonal effect, cholesterol levels being some 3-5% higher in winter than in summer. This was independent of seasonal changes in body mass. Mean monthly cholesterol levels were negatively correlated with mean monthly air temperatures (r = -0.60 to -0.71). The observed seasonal differences were larger than the inter-assay coefficient of variation for the cholesterol determination method employed and have important consequences for long-term epidemiological or intervention studies.