Population studies show a relation of alcohol drinking to an increased risk of breast cancer (BrCa). Aiming to investigate uncertainties about a risk threshold, the role of beverage type and interactions with other BrCa predictors, we performed a cohort study among 70,033 women, 2,829 of whom developed BrCa. Using Cox proportional hazards models with 8 covariates, the following relative risks (95% confidence intervals) for BrCa versus lifelong abstainers were found: 1.08 (0.95-1.22) at <1 drink per day, 1.21 (1.05-1.40, p=0.01) at 1-2 drinks daily and 1.38 (1.13-1.68, p=0.002) at > or = 3 drinks daily. Increased BrCa risk was concentrated in women with oestrogen receptor positive tumours with no major disparity related to choice of wine, liquor, beer or type of wine (red, white, etc). We conclude that with a threshold below 1-2 drinks daily, a hormone-related mechanism mediates a relation of alcohol drinking to an increased BrCa risk.