A matched case-control study was done on 175 pathologically proven new cases of female breast cancer from National Taiwan University Hospital from February 1993 to June 1994 and 457 hospital controls individually matched for age and date of admission. Height and weight during one year before the first full-term pregnancy and one year before admission were collected from a questionnaire interview. Body mass index (BMI) (kg/m2) was derived accordingly. Demographic and reproductive characteristics, and family history of breast cancer were collected and adjusted as potential confounders. Conditional multiple logistic regression analysis was used to estimate the effect for each risk factor regarding body size of different ages. After adjustment for potential confounders, higher weight and BMI during one year before first full-term pregnancy were associated with a moderately lower risk of breast cancer especially in the premenopausal group. The heaviest group (> or = 65 kg) during one year before admission also had a moderately lower breast cancer risk. Height did not show any significant association with breast cancer risk. Weight and BMI gain were associated with an elevated breast cancer risk in the postmenopausal group but with wide confidence limits. The results of this study imply that adipose tissue may interfere with ovarian endocrine in the younger age group, and that there is increased hormonal production after obvious weight gain.