In order to investigate the value of ras oncogene expression as a prognostic indicator in rectal adenocarcinoma, we evaluated the level of ras gene protein product (p21) in the available material of 149 consecutive Dukes' B and C specimens resected at our institution between 1965 and 1981. Five year follow-up was available in all patients. Pathology slides and archival paraffin blocks were retrieved for confirmation of the original diagnosis, study of histopathological features, and measurement of p21 content. P21 titers were obtained using the RAP-5 monoclonal antibody in a semiquantitative immunohistochemical assay. Titer was expressed as the highest dilution of antibody giving definitive staining using the avidin-biotin peroxidase method. The analysis indicated that the group of tumors with high (greater than or equal to 1:40,000) p21 titers had a worse 5-year survival (43.9 versus 64.3%, P less than 0.02), higher incidence of distant metastases (51.8 versus 23.2%, P less than 0.001), and more advanced Dukes' stage (53.7 versus 35.7% incidence of Dukes' C stage, P less than 0.04) than tumors with low (less than 1:40,000) titers. Multivariate analysis also demonstrated that the influence of the level of ras gene protein product on survival was independent of Dukes' stage. We conclude that detection of high levels of ras oncogene protein product indicates a group of tumors with a more aggressive behavior, characterized by a higher percentage of distant recurrences and worse prognosis. These findings suggest that measurement of p21 may become clinically important in identifying patients at high risk of recurrent disease.