The association between total serum cholesterol and triglycerides and the incidence rate of breast cancer has been examined in 242 incident cases of breast cancer that developed among 24,329 Norwegian women during 11-14 years of follow-up. At the time of lipid measurement they were between 35 and 51 years of age. There was an inverse relation between serum cholesterol and risk of breast cancer which was confined to women diagnosed before the age of 51. The incidence rate ratio was 0.53 (95% confidence limits, 0.32 and 0.88) for women in the highest quartile of serum cholesterol (mean = 8.52 mM = 329 mg/100 ml) compared to women in the lowest quartile (mean = 5.28 mM = 204 mg/100 ml), and the relation displayed a negative trend over quartiles of cholesterol (chi 2 for trend = 3.94, P = 0.05). No association with cholesterol was found among cases diagnosed during the initial 2 years of follow-up, whereas a strong inverse relation was observed among cases that were diagnosed during the following 4 years (chi 2 trend = 12.6, P less than 0.001). For serum triglycerides there was an overall negative, but not statistically significant, association with breast cancer incidence, which was weakened after further adjustment for body mass index and serum cholesterol. The lack of information on reproductive factors associated with breast cancer risk is a limitation of this study, since potential confounding with the results cannot be excluded. We conclude that there is an inverse relation between serum cholesterol and breast cancer risk among women diagnosed before the age of 51 years. The findings indicate that the negative association cannot easily be attributed to a preclinical effect of the cancer.