Two hundred fifty-seven eligible patients with stage I, IIA "high risk" ovarian carcinoma and IIB, IIIO (disease confined to pelvis), were randomized to either total abdominal radiotherapy (arm A) 2,250 rad in 20 fractions (107 patients), melphalan (arm B) 8 mg/m2/d X 4 every 4 weeks X 18 courses (106 patients), or intraperitoneal chromic phosphate (arm C) 10 to 20 mCi (44 patients). All patients were initially treated with pelvic radiotherapy; arm A, 2,250 rad in ten fractions; and arms B and C, 4,500 rad in 20 fractions. Entry to arm C was discontinued early because of toxicity. In a multifactor analysis using proportional hazards models, no significant difference in survival was observed although there was a marginally significant difference in disease-free survival (P = .015) with arm B being superior to arm A. Stage (P less than .0001), grade (P less than .0001), and histology (P less than .008) were predictors of survival in the multifactor analysis. Performance status, age, and residual disease were significant predictors in the single factor analysis but were not predictive when correction was made for the effects of stage, grade, and histology. Five-year survival rates are 62% for arm A, 61% for arm B, and 66% for arm C. Median duration of follow-up is 8 years. Long-term complications of radiotherapy were seen in 19 patients on arm A, 11 on arm B, and 11 on arm C. Four patients who had received melphalan developed either a myelodysplastic syndrome or acute leukemia. Violations in covering the whole abdominal target volume were correlated with survival.