Methods We sent a questionnaire to 1431 women younger than age 75 years who had breast cancer and were listed on the population-based British Columbia cancer registry between June 1, 1988, and June 30, 1989. We also sent questionnaires to 1502 age-matched controls, randomly selected from the 1989 provincial voters list. We obtained information on all known and suspected risk factors for breast cancer, and on lifetime smoking, alcohol consumption, and occupational history. We assessed the effect of smoking separately for premenopausal and postmenopausal women, adjusting for confounding variables.

Findings 318 premenopausal women and 340 controls replied. Risk of breast cancer was significantly increased (adjusted odds ratio 1.69, 95% CI 1.13-2.51) in women who had been pregnant and who started to smoke within 5 years of menarche, and in nulliparous women who smoked 20 cigarettes daily or more (7.08, 1.63-30.8) and had smoked for 20 cumulative pack-years or more (7.48, 1.59-35.2). Postmenopausal women (700 breast cancer and 685 controls) whose body-mass index increased from age 18 to current and who started to smoke after a first fullterm pregnancy had a significantly reduced risk of breast cancer (0.49, 0.27-0.89).

Interpretation Our results suggest that cigarette smoke exerts a dual action on the breast, with different effects in premenopausal and postmenopausal women. Our observations reinforce the importance of smoking prevention, especially in early adolescence, and draw attention to the timing of exposure in relation to susceptibility and refractory windows in the design of studies to investigate associations between environmental carcinogens or putative endocrine disruptors and risk of breast cancer.