A collaborative, hospital-based case-control study was conducted at 12 participating centres in 10 countries. Based on data from personal interviews of 2,116 women with newly diagnosed breast cancer and 12,077 controls, the relative risk of breast cancer in women who ever used oral contraceptives was estimated to be 1.15 (1.02, 1.29). Estimated values of this relative risk based on data from three developed and seven developing countries were 1.07 (0.91, 1.26) and 1.24 (1.05, 1.47) respectively; these estimates are not significantly different (P = 0.22). Estimates for women under and over age 35 were 1.26 (0.95, 1.66) and 1.12 (0.98, 1.27), respectively, and these estimates are also not significantly different (P = 0.38). Risk was highest in recent and current users and declined with time since last use regardless of use. Risk did not increase with duration of use after stratifying on time since last use. Risk did not increase significantly with increasing duration of use before age 25 or before a first live birth. However, a relative risk of 1.5 that was of borderline statistical significance was observed in women who used oral contraceptives for more than 2 years before age 25. No single source of bias or confounding was identified that could explain the small increases in risk that were observed. Chance alone is also an unlikely explanation. The results could be due to a combination of chance and potential sources of bias, or they could represent a weak causal relationship.