We examined the relation between menstrual irregularity and breast cancer mortality in the American Cancer Society Cancer Prevention Study I. In 1959, 428,653 women age 30 years and older completed a questionnaire on possible cancer risk factors, including their recollection of their menstrual pattern at age 20 years. We used proportional hazards analysis to estimate rate ratios, based on 2,226 breast cancer deaths over 13 years of follow-up. Women who reported having irregular menstruation at age 20 years were at reduced risk for breast cancer mortality (rate ratio = 0.84; 95% confidence interval = 0.74-0.96). The relation was neither confounded nor modified by other breast cancer risk factors.