P53 tumor suppressor gene protein immunostaining was evaluated in the primary tumor of adult testicular germ cell cancer to assess if P53 expression would serve as a clinically useful tumor marker. Representative archival tissues from 152 orchiectomy specimens were studied for P53 immunohistochemistry. Seminoma and nonseminomatous germ cell tumor constituents revealed P53 expression via immunohistochemistry in 90% and 94% of the cases, respectively. For seminoma, there was a trend toward decreased P53 expression with advancing stage. For nonseminomatous germ cell tumor, although all cellular components showed variable P53 expression, P53 expression in embryonal carcinoma constituents increased among stages of disease. A third of pathological stage I cancer patients exhibited 2+ or greater P53-embryonal staining compared with 61% with stage II (p = 0.0670) and 67% with stage III (p = 0.0815) disease, respectively (Kruskal-Wallis, 2-sided test). As a secondary objective, we wanted to determine if P53 immunohistochemistry would be useful to predict occult disease in clinical stage I nonseminomatous germ cell tumor. This group was studied for P53-embryonal immunohistochemistry, the presence of vascular invasion and the quantitative determination of percentage of embryonal carcinoma in the primary tumor in a multivariate fashion to assess if these tests could be clinically useful to predict occult disease. Degree of P53 immunostaining of the embryonal component in the primary tumor was statistically greater for stage II by univariate logistic regression analysis (p = 0.0362). Similarly, the per cent embryonal cancer (p = 0.0002) and vascular invasion (p = 0.0005) were highly significant as predictors of occult stage II disease via the univariate testing. By multivariate logistic regression analysis, the model consisting of per cent embryonal cancer and vascular invasion provided the best prediction of occult disease in the clinical stage I cohort. In addition, this model had the highest sensitivity and specificity of all multivariate models considered. The addition of P53-embryonal staining did not improve predictability nor sensitivity/specificity. The P53 tumor suppressor gene protein is expressed to some degree in most testicular germ cell tumors and degree of staining/expression varies according to stage of disease. From the standpoint of a clinically useful primary tumor risk factor for predicting occult disease, vascular invasion by the tumor and percentage of embryonal carcinoma component in the tumor are more useful than P53 immunohistochemistry.