A population-based series of 649 unselected incident cases of ovarian cancer diagnosed in Ontario, Canada, during 1995-96 was screened for germline mutations in BRCA1 and BRCA2. We specifically tested for 11 of the most commonly reported mutations in the two genes. Then, cases were assessed with the protein-truncation test (PTT) for exon 11 of BRCA1, with denaturing gradient gel electrophoresis for the remainder of BRCA1, and with PTT for exons 10 and 11 of BRCA2. No mutations were found in all 134 women with tumors of borderline histology. Among the 515 women with invasive cancers, we identified 60 mutations, 39 in BRCA1 and 21 in BRCA2. The total mutation frequency among women with invasive cancers, 11.7% (95% confidence interval [95%CI] 9.2%-14.8%), is higher than previous estimates. Hereditary ovarian cancers diagnosed at age <50 years were mostly (83%) due to BRCA1, whereas the majority (60%) of those diagnosed at age >60 years were due to BRCA2. Mutations were found in 19% of women reporting first-degree relatives with breast or ovarian cancer and in 6.5% of women with no affected first-degree relatives. Risks of ovarian, breast, and stomach cancers and leukemias/lymphomas were increased nine-, five-, six- and threefold, respectively, among first-degree relatives of cases carrying BRCA1 mutations, compared with relatives of noncarriers, and risk of colorectal cancer was increased threefold for relatives of cases carrying BRCA2 mutations. For carriers of BRCA1 mutations, the estimated penetrance by age 80 years was 36% for ovarian cancer and 68% for breast cancer. In breast-cancer risk for first-degree relatives, there was a strong trend according to mutation location along the coding sequence of BRCA1, with little evidence of increased risk for mutations in the 5' fifth, but 8.8-fold increased risk for mutations in the 3' fifth (95%CI 3.6-22.0), corresponding to a carrier penetrance of essentially 100%. Ovarian, colorectal, stomach, pancreatic, and prostate cancer occurred among first-degree relatives of carriers of BRCA2 mutations only when mutations were in the ovarian cancer-cluster region (OCCR) of exon 11, whereas an excess of breast cancer was seen when mutations were outside the OCCR. For cancers of all sites combined, the estimated penetrance of BRCA2 mutations was greater for males than for females, 53% versus 38%. Past studies may have underestimated the contribution of BRCA2 to ovarian cancer, because mutations in this gene cause predominantly late-onset cancer, and previous work has focused more on early-onset disease. If confirmed in future studies, the trend in breast-cancer penetrance, according to mutation location along the BRCA1 coding sequence, may have significant impact on treatment decisions for carriers of BRCA1-mutations. As well, BRCA2 mutations may prove to be a greater cause of cancer in male carriers than previously has been thought.