The effect of regularity and length of the menstrual cycle on breast cancer risk was studied prospectively in 78 cases and 383 age-matched controls who participated in a breast cancer screening programme, the DOM-project, in Utrecht, the Netherlands. Before entering the screening programme when they were aged 41-46, the women kept a menstrual calendar during at least three consecutive cycles. Cycles were considered to be irregular if any of three cycles was shorter than 21 days or longer than 35 days and/or if variation between cycle lengths was more than five days. Women with irregular cycles had a significantly reduced risk of breast cancer (odds ratio = 0.44; 95% confidence interval 0.22-0.86) after adjustment for age at menarche, age at first birth, parity, Quetelet's index and family history of breast cancer. Among regularly menstruating women, long cycles (28 days or more) were not significantly associated with increased risk of breast cancer (odds ratio 1.17; 95% confidence interval 0.66-2.09). To the extent that irregular menstrual cycles reflect anovulatory cycles, our findings support the hypothesis that the cumulative number of regular ovulatory cycles increases breast cancer risk.