The majority of patients have 'good risk' nonseminomatous germ cell tumors, that is, they are likely to obtain an initial complete response to chemotherapy. Trials in these patients attempt to reduce the toxicity of chemotherapy. The minority of patients have refractory disease ('poor risk' nonseminomatous germ cell tumors) and the primary objective in this population is to improve the proportion of complete response. Chemotherapy trials for the poor risk group are generally associated with greater toxicity. Clinical factors reported to impact on the prognosis of nonseminomatous germ cell tumor patients include the pretreatment level of serum tumor markers (AFP, HCG, LDH), extern or bulk of disease, the pathology and the site of the primary lesion. No consensus currently exists on which factors are most important in determining prognosis and how they should be used to allocate patients to good and poor risk nonseminomatous germ cell tumors trials. The data support the concept that biologic markers and extent or bulk of disease together are more predictive of response and survival than either one alone and their combined use results in a smaller percent of the germ cell tumor population allocated to poor risk trials. This minimizes exposure of good risk patients to toxic regimens and still maintains a high response rate in the good risk population. The current eligibility criteria for good and poor risk trials in use directly affects the reported response proportion and survival. In view of these differences, good and poor risk trials should be randomized against a comparative control population and nonrandomized trials comparing treatment results with those of other studies or historical controls should be viewed with extreme caution.