Relatives of postmenopausal breast cancer patients have a greater risk of breast cancer than do women with no family history of the disease. However, no study has previously examined the contribution of other breast cancer risk factors to risk in these relatives, thus precluding the ability to discern between shared environment or genes as the cause of familial aggregation. We conducted a nested case-control study of 251 index case families and 234 index control families. Incident breast cancer cases occurring in 1988-1989 and cancer-free controls were selected from a prospective cohort study. A pedigree development form was mailed to identify female relatives, who were then mailed a health history questionnaire. Index cases were significantly more likely than index controls to have at least one first-degree female relative with breast cancer [odds ratio = 1.62; 95% confidence interval = 1.03-2.54] after controlling for the total number of first-degree female relatives. After excluding the index cases and controls, logistic regression models were used to predict the odds of breast cancer in sisters (n = 72 cases). Sisters of index cases had slightly, although not statistically significant, elevated risk of both early-onset (age < 50 years; odds ratio = 1.34; 95% confidence interval = 0.54-3.30) and late-onset (age > or = 50 years; odds ratio = 1.56; 95% confidence interval = 0.87-2.79) breast cancer. Adjustment for nongenetic risk factors increased the odds ratio to 1.41 and 1.81, respectively. These data suggest that risk to sisters of postmenopausal breast cancer cases may be increased for both early- and late-onset breast cancer and this risk is apparently independent of measured risk factors.