The prognostic value of subjective histological and morphometric grading was studied in 75 primary ovarian carcinomas. Histological grading methods recommended by Czernobilsky and by Russell and the morphometric method of Baak and co-workers were compared in a two-observer system. The 5-year survival could be correctly predicted in about two-thirds of the patients with all three methods. When mitotic counting (volume corrected mitotic index, M/V-index) was compared with the above grading methods by using a receiver operating characteristic curve) the M/V-index was generally superior in its prognostic power regardless of the sensitivity/specificity level chosen. The morphometric grading method and the grading method based on the M/V index were also shown to be readily reproducible.