Between 1978 and 1982, 180 patients from Indiana University (Indianapolis) were entered on the Southeastern Cancer Study Group (SECSG) protocol 78 GU 240, a randomized comparison of cisplatin, vinblastine, and bleomycin (PVB) v PVB plus doxorubicin induction chemotherapy regimens, with a second randomization to maintenance vinblastine v no further therapy. One hundred forty-eight of these patients obtained a favorable response to chemotherapy, defined as a complete response (CR) or a surgical resection of teratoma. The prognostic significance of various patient characteristics was investigated using the logistic regression model. Two classifications for the extent of disease were considered: the Indiana staging system and the M.D. Anderson (MDA) staging system. The Indiana staging system had the greater prognostic significance. This staging system allowed the population to be split into three groups (minimal, moderate, advanced disease) in which the observed proportions of favorable responders were 99%, 90%, and 58%, respectively. Within the advanced group, the number of elevated tumor markers subdivided these patients into three groups, with the observed proportions of favorable responders being 73%, 65%, and 45%. The Indiana and MDA staging systems were subsequently prospectively used in SECSG protocol GU 81 332, a study randomizing patients to remission induction therapy with PVB v cisplatin, VP-16, and bleomycin. The prognostic value of the Indiana staging system was prospectively validated in this study.