One hundred forty-three incidental carcinomas from 500 step-sectioned prostates were classified independently by two pathologists according to the Gleason system. The area involved by each tumor was measured by photographic techniques. Observer agreement was 94 per cent when tumors were divided into groups with "pattern scores" 4 or less (less than or equal to 4) and greater than 4 (greater than 4). About 25 per cent of tumors in both races had pattern scores 4 or less (less than or equal to 4). Tumors with pattern scores greater than 4 (greater than 4) were, on the average, significantly larger than those with pattern scores 4 or less (less than or equal to 4) (p less than 0.001). We therefore believe that the Gleason histologic classification system is reproducible and that it discriminates tumors with different biological behavior.