Predictive factors for ovarian metastases were analyzed in 313 autopsy cases of women dying of malignancy. Genital, hematopoietic, brain tumors, and double primaries were excluded. Ovarian metastases were found in 60 of 313 (19.2%) cases. The primary sites of origin were stomach (18 cases), breast (13), pancreas (7), bile duct (7), large intestine (4), kidney (3), liver (2), lung (1), and other (5). The percentages of ovarian metastasis in each primary site were 50% (3/6) in the kidney, 38.3% (18/47) in the stomach, 30.8% (4/13) in the large intestine, 22.9% (13/57) in the breast, 22.6% (7/31) in the bile duct, 19.4% (7/36) in the pancreas, 5.4% (2/37) in the liver, and 3.4% (1/29) in the lung. Seventy-five percent (75%) of ovaries with metastasis were 5 cm or less in the greatest dimension. Seventy-eight percent (78%) of ovarian metastases were bilateral. Fifty-one percent (51%) of the involved ovaries were cystic. Twenty-four percent (24%) of the metastasis were occult in which microscopic disease was present in normal appearing ovaries. The mean age of death in the cases with ovarian metastasis was 53.8 +/- 16.8 years. This was significantly younger than the mean age without ovarian metastasis that mainly was attributable to the younger ages for breast and bile duct cancer. The number of other extraovarian metastatic sites for positive ovarian cases was significantly greater than for the negative ovarian cases (7.6 +/- 2.4 and 5.8 +/- 2.5, respectively). This difference was dominant in gastric and bile duct cancers. The other metastatic sites that accompanied ovarian metastases were the adrenal gland in gastric cancer, the spleen and small intestine in breast cancer, and the adrenal gland and colon in bile duct cancer. The frequency of peritoneal dissemination was essentially the same between cases with positive and negative ovarian metastasis. Scirrhous adenocarcinoma of the stomach more commonly metastasized to the ovary than any other stomach histological type. These findings suggest that although the impact upon outcome of bilateral salpingo-ophorectomy at the initial operation is unknown, the discovery of ovarian metastases in a large proportion of patients may dramatically alter initial management in patients otherwise thought to have only local disease.