We report the results of a case-control study of oral contraceptive use and breast cancer conducted in London, Oxford and Edinburgh between 1980 and 1984. One thousand one hundred and twenty-five women aged 16-64 years with newly diagnosed breast cancer and a like number of matched controls were interviewed and asked about their past due use of oral contraceptives (OCs). Among women aged 45 years or more at diagnosis there was no evidence of an association between OC use and breast cancer. Among the 351 pairs of women aged under 45 years at diagnosis there was a significantly elevated risk associated with increasing duration of use before first full term pregnancy (relative risk for 4+ years use versus never use = 2.6, 95% confidence limits, 1.3-5.4). Since this result is at variance with the findings in some other studies we have investigated the nature of this association with particular emphasis on possible bias, pill type and a latent effect.