During 1968-1980, 1176 women aged 16-50 years with newly diagnosed breast cancer and a like number of matched controls were interviewed at 9 teaching hospitals in London and Oxford and asked about their use of oral contraceptives. The results were reassuring. A few statistically significant differences in oral contraceptive use were found between the breast cancer and control groups, but the data were subdivided in many ways so that some "significant" differences would have been expected through the play of chance alone. Certainly no patterns of risk emerged which would suggest that any of the associations were causal. It must be stressed, however, that the data are still sparse in some important subcategories--for example, only small numbers of both cases and controls had prolonged oral contraceptive use before their first term pregnancy. For this reason, it is important that information on the possible relationship between pill use and breast cancer should continue to be collected. Women who had never used oral contraceptives presented with appreciably more advanced tumours than those who had been using oral contraceptives during the year before detection of cancer, while past users were in an intermediate position. These differences in staging were reflected in the pattern of survival. Possible explanations for these observations include "surveillance bias" among oral contraceptive users leading to earlier diagnosis and a beneficial biological effect of oral contraceptives on tumour growth and spread. Women with breast cancer reported never having used any method of contraception and heavy cigarette smoking (greater than or equal to 15 per day) significantly less often than controls. We could find no obvious explanation for the former observation, but suspect that the latter reflects the unrepresentative smoking habits of our hospital controls rather than a protective effect of smoking against breast cancer.