In order to evaluate possible differences in the typing and grading of ovarian tumors, four different gynecologic pathologists independently evaluated slides from 198 patients. Histologic typing was done using the WHO recommendations, and the tumors were graded as benign, borderline or malignant (well, moderately or poorly differentiated). All of the slides were assessed twice by each pathologist (with approximately a 12-month time interval) in a "blind" fashion (i.e., without any knowledge about stage, treatment and clinical outcome). In addition to assessing interobserver agreement, the intraobserver consistency was evaluated by comparing the first and second assessments of the same pathologist. Histologic grading showed the best correlation between observers' results; however, the level of agreement was low. Complete agreement was rare, both in the first and second assessments (18.7% and 32.8%, respectively). Complete disagreement was rare, but occurred, both in the first and second assessments. Agreement in histologic typing was rather good after panel discussions; further analysis of the data revealed that the lack of agreement could not be attributed to any one of the pathologists in particular. The intraobserver agreement in grading and typing the same tumor differed. The number of cases with more than one grade difference between the first and second assessments varied from 0.5% to 3.2% between the different pathologists. Complete consistency was also varied: 87%, 78%, 64% and 62%. Histologic typing in general was less consistent, although some pathologists were better than others. The results indicate the need for objective criteria for grading ovarian tumors.