Grading of ovarian carcinomas can have important implications for therapeutic decisions, in particular in International Federation of Gynecology and Obstetrics (FIGO) stage I. However, there are no universally accepted grading guidelines for this type of cancer. We applied the grading system suggested by Shimizu et al. (1) to a series of ovarian carcinomas from a single institution to evaluate its prognostic significance in relation to other predictive factors. One hundred ninety-two cases of ovarian carcinomas were studied, including all major histologic types. The histologic slides were evaluated with regard to architecture (glandular = 1 point, papillary = 2 points, solid = 3 points), nuclear pleomorphism (nuclear variability < or = 2:1 = 1 point, intermediate nuclei = 2 points, nuclear variability > or = 4:1 = 3 points), and mitotic activity per 10 high-power fields using objective 40x, ocular 10x/20 (0-7 mitoses = 1 point, 8-18 mitoses = 2 points, > or = 9 mitoses = 3 points). Carcinomas with a total score of 3-5 points were designated as grade I, 6-7 points were designated as grade II, and 8-9 points were designated as grade III. Kaplan-Meier curves showed the following 5-year survival rates: grade I (n = 42) 88%, grade II (n = 98) 60%, grade III (n = 52) 38% (p < 0.0001); stage I 90%, stage II 60%, stage III 38%, stage IV 10% (p < 0.0001); residual disease < or = 2 cm 67% and > or = 2 cm 27% (p < 0.0001). Multivariate Cox analysis revealed that grade (p < 0.0002), as well as nuclear pleomorphism alone (p < 0.0001), both provided statistically significant independent prognostic information. Our observations showed that the grading system used can be easily applied to all histologic types of ovarian carcinomas yielding prognostically relevant information and can be incorporated into routine diagnostic practice.