BACKGROUND Previous studies of BRCA1 mutation prevalence have been based on high-risk groups, yielding estimates that do not reflect the experience of the general population of US patients with breast cancer. OBJECTIVE To determine prevalence of known disease-related mutations and other variants in BRCA1 and how it differs by race, age at diagnosis, and family history status in a population-based sample of white and black patients with breast cancer unselected for family history. METHODS Case-control study. METHODS A 24-county area of central and eastern North Carolina. METHODS Cases were women aged 20 to 74 years diagnosed as having a first invasive breast cancer between May 1993 and June 1996. Controls were frequency matched to cases by 5-year age range and race. The first 211 cases and 188 controls regardless of race and the subsequent 99 cases and 108 controls of African American ancestry are included in this report. METHODS Germline variants at any site in the coding sequence, splice junctions, 5' untranslated region, or 3' untranslated region of the BRCA1 gene were analyzed in cases, and selected variants were analyzed in controls. Screening was performed using multiplex single-strand conformation analysis, with all potential variants confirmed using genomic sequencing. RESULTS Three of 211 patients with breast cancer had disease-related variants at BRCA1, all of which were protein-truncating mutations. After adjustment for sampling probabilities, the proportion of patients with breast cancer with disease-related variants was 3.3% (95% confidence interval, 0%-7.2%) in white women and 0% in black women. Young age at diagnosis alone did not predict BRCA1 carrier status in this population. In white women, prevalence of inherited mutation was 23% for cases with family history of ovarian cancer, 13% for cases from families with at least 4 cases of breast cancer with or without ovarian cancer, and 33% for cases from families with both breast and ovarian cancer and at least 4 affected relatives. Because these results are based on few families at the highest levels of risk, confidence intervals around these estimates are wide. An additional 5 patients had rare missense mutations or a single amino acid deletion, the biological significance of which is unknown. In black women, a variant in the 3' untranslated region was statistically significantly more common in cases than in controls. CONCLUSIONS These data suggest that in the general US population, widespread screening of BRCA1 is not warranted. In contrast, BRCA1 mutations are sufficiently frequent in families with both breast and ovarian cancer, or at least 4 cases of breast cancer (at any age), that genotyping might be considered. The emerging picture of BRCA1 population genetics involves complex interactions of family history, age, and genetic ancestry, all of which should be taken into account when considering testing or interpreting results.