Epidemiological studies have implicated smoking as a possible risk factor in the etiology of breast cancer, yet the evidence is not conclusive. We conducted meta-analyses of peer-reviewed studies published between 1984 and 2001 to assess the relation between smoking and breast cancer. The studies were located by searching the MEDLINE (1966 to 2001) and Cancer Abstracts (1980 to 2001) databases. Combined estimators of relative risk (RR) were calculated using fixed and random effect models. The combined RR for ever smokers was 1.10 (95% CI = 1.02-1.18). The association was stronger in premenopausal cases (RR = 1.21, CI = 1.08-1.36). The dose-response trend was significant but weak for the number of cigarettes smoked per day and for the duration of smoking. Early age at the start of smoking was associated with elevated risk (RR = 1.14, CI = 1.06-1.23). Our results suggest that smoking is a weak risk factor for breast cancer and the risk is higher in the premenopausal period and in those who started smoking at an early age.