One hundred and seventy-five pathologically confirmed new cases of female breast cancer in National Taiwan University Hospital were selected as the case group. Four hundred and fifty-seven inpatient controls were individually matched for each case by age and date of admission. Information of reproductive factors and potential confounders were collected from questionnaire interview and medical records. Menstrual cycles before and after first fuIl-term pregnancy were estimated from the questionnaire. Conditional logistic regression was used to estimate the effects of reproductive risk factors and numbers of menstrual cycles before and after first full-term pregnancy. After adjusting for educational level, body mass index, family history of breast cancer and oral contraceptive use, age at menarche and menopause only showed borderline significance. Age at first full-term birth predicted breast cancer risk better. Parity was the most important protective factor. Number of menstrual cycles before first full-term pregnancy was also positively associated with breast cancer risk with linear trends.