Between 1979 and 1984, 53 patients received whole abdominal irradiation in a curative salvage effort for residual (32 patients) or recurrent (21 patients) epithelial ovarian cancer after combination chemotherapy (cisplatin-based in 48 patients). Residual cancer less than or equal to 2 cm in diameter was confirmed at operation in all patients before irradiation consisting of 2,550 to 3,000 rad to the whole abdomen with partial liver/kidney shielding and boosting of the dose to the diaphragmatic/paraaortic nodal regions and pelvis to approximately 4,000 and 5,000 rad, respectively. Twelve patients (23%) did not complete therapy as a result of hematologic intolerance. Actuarial overall and disease-free survival at 3 years are 35% and 30%, respectively, with follow-up for disease-free patients ranging from 30 to 79 months (median, 43 months). Twenty-seven of 36 relapses (75%) occurred within the irradiated abdomen alone. At 3 years, 70% of patients with well- or moderately-differentiated tumors were disease-free v 10% of those with poorly differentiated tumors (P less than .001). Among prognostic factors evaluated, including grade, initial residual disease before chemotherapy, residual disease at time of irradiation, age, chemotherapy response v progression, and completion of irradiation, only grade and initial residual disease before chemotherapy were statistically significant in multivariate analysis (both P less than .01). Patients with the combination of high-grade tumor, initial residual disease greater than 2 cm before chemotherapy, and macroscopic disease after "second-look" laparotomy do not benefit from irradiation. Eleven patients (21%) developed an apparent treatment-related bowel obstruction after completion of irradiation. Selected subsets of patients do well; however, the role of irradiation in this setting can be confirmed only with randomized clinical study.