Two case-control studies of Canadian women aged 40-59 years are reported investigating the relation of cigarette smoking with initial visit (prevalent) and subsequent visit (incident) breast cancer detection, respectively, within the Canadian National Breast Screening Study. The analysis of prevalent breast cancer (1982-1985), which involved 254 cases and 762 controls, showed no evidence of an elevated risk for women with a history of cigarette smoking, with odds ratios of 0.9 (95% confidence interval (Cl) 0.6-1.5) and 1.2 (95% Cl 0.8-1.8) in premenopausal and postmenopausal subjects, respectively. Similarly, in the incident breast cancer analysis (1981-1987) based on 317 cases and 951 controls, women with a history of cigarette smoking had odds ratios of 1.2 (95% Cl 0.8-2.0) and 1.2 (95% Cl 0.9-1.7) in the premenopausal and postmenopausal categories, respectively. No evidence of dose response or of elevated risk in ex-smokers or current smokers was found in either study. These results persisted despite adjustment for several important variables. The present data demonstrate no association between smoking and prevalent or incident breast cancer in either menopausal category, contradicting the authors' previous finding of a positive association with premenopausal prevalent disease earlier in this screening study. The relation of smoking and breast cancer remains controversial. Further study is required to determine whether an association truly does not exist or whether smoking might have both protective and harmful effects that are mediated through different pathways, thus accounting for the paradoxical findings in the literature to date.