The association between coffee consumption and the incidence rate of breast cancer has been analysed in 152 incident cases of breast cancer that developed among 14,593 Norwegian women during a mean follow-up of 12 years. At the time of inquiry they were between 35 and 51 years of age, and at the end of follow-up between 46 and 63. There was an overall weak negative association between daily intake of coffee and risk of breast cancer, which was not statistically significant. However, the association with coffee varied, depending on the body mass index (BMI) of the women. In the lean (Quetelet less than 24; population mean) there was an inverse relation between coffee intake and risk of breast cancer (chi 2 trend = 5.07, P = 0.02). In this group, women who reported drinking 5 cups or more per day had an age-adjusted IRR of 0.5 (95% confidence intervals, 0.3 and 0.9) compared to women who had 2 cups or less. In women with Quetelet's index equal to or greater than 24 there was a positive relation between coffee intake and breast cancer risk (chi 2 trend = 2.33, P = 0.13), where the corresponding age-adjusted IRR was 2.1 (95% confidence intervals, 0.8 and 5.2). This interaction effect between coffee intake and BMI was statistically significant (chi 2 interaction = 10.2, 3 d.f., P = 0.02). In summary, the results of this study suggest that coffee consumption reduces the risk of breast cancer in lean women, whereas coffee might have the opposite effect in relatively obese women.