Paraffin-embedded tissue samples from 256 patients who received primary treatment (surgical staging, reduction of tumor size, and adjuvant therapy based on surgical and pathologic risk factors) for endometrial carcinoma at the Mayo Clinic between 1979 and 1983 were analyzed by flow cytometry to determine DNA ploidy characteristics. Diploid patterns constituted 78% of the cases, whereas aneuploid and tetraploid patterns accounted for 17% and 5%, respectively. Only 10% of patients with diploid tumors had a relapse in comparison with 39% of those with nondiploid lesions (34% with aneuploid; 58% with tetraploid). Significant differences (P less than 0.001) were noted in estimated 4-year progression-free survivals--88% for patients with diploid and 57% for those with nondiploid tumors. Stage, grade, depth of myometrial invasion, histologic subtype, peritoneal cytology, and DNA ploidy all demonstrated independent prognostic significance (P less than 0.001) in this study population. When subjected to multivariate analysis, however, grade and depth of myometrial penetration failed to retain prognostic significance (P greater than 0.15) and surgical stage was marginally significant (P = 0.05), whereas histologic subtype and DNA ploidy maintained significant predictive powers (P less than 0.001 and P less than 0.01, respectively). We conclude that DNA ploidy is a major objective prognostic factor and therapeutic determinant for endometrial carcinoma.