Few data exist on the extent to which the differences in breast cancer risk between "racial-ethnic" groups in the United States (US) are "explained" by differences in their distribution of risk factors. We have determined this for African-American (AA), native Hawaiian (NH), Japanese-American (JA), Latina-US-born (L-US), Latina-non-US-born (L-NUS), and white (W) women using prospective incidence data on 88,712 postmenopausal women recruited in 1993-1996. We identified 1,757 incident breast cancer cases through 1999 among these women (1,116 cases after excluding women with a simple hysterectomy or missing risk factor data). Data were available on seven "known" risk factors: ages at menarche and first birth; parity; age at and type of menopause; weight; hormone replacement therapy use; and alcohol consumption. The relative risks (RRs) of breast cancer (with the RR in Ws set to 1.0) for the groups were as follows: W = 1.0; AA = 0.78; NH = 1.33; JA = 0.99; L-US = 0.77; and L-NUS = 0.60. After adjustment for the risk factors, the RRs were as follows: W = 1.0; AA = 0.98; NH = 1.65; JA = 1.11; L-US = 0.95; and L-NUSB = 0.84. The slightly greater risk of the JAs compared with the Ws is in sharp contrast to the very low breast cancer rates that were observed in "traditional" Japanese women and in early Japanese migrants. The adjusted RR of NHs is 65% greater than that of Ws, and that of migrant Latinas is 16% lower than that of Ws. Elucidating the causes of the high rates in NHs is now a major focus of our efforts.