Results from studies evaluating the relationship between cigarette smoking and breast cancer have been inconsistent. Though most studies have found that smoking does not alter risk, others have observed both increased and decreased risks associated with smoking. The reasons for these inconsistencies are unclear, but they may be related to differences in study populations, designs, and exposure definitions. In particular, this relationship may vary by age, and few studies have focused on older women many of whom have smoked for very long durations. We conducted a population-based case-control study (975 cases/1007 controls) of women 65-79 years of age in western Washington State. Women who were current smokers, smoked for > or =40 years, had > or =11 pack-years of lifetime smoking, or started smoking before their first full-term birth each had 30-40% elevated risks of breast cancer (p < 0.05). Recency, length, and intensity of smoking are all associated with modest increased risks of breast cancer. A further understanding of the timing of smoking, and its interaction with other factors, may enhance our knowledge of whether and by what mechanisms smoking alters breast cancer risk.