In recent years, the prognosis of non-seminomatous germ-cell tumors has been greatly improved by the use of novel chemotherapeutic regimens including platinum derivatives. However, the prognosis remains poor for a certain proportion of these patients. We have therefore developed an intensive chemotherapy protocol (PEC) followed by bone marrow autografting: cisplatin (40 mg/m2/day x 5 d), etoposide (350 mg/m2/day x 5 d), cyclophosphamide (1,600 mg/m2/day x 4 d). Forty-four poor-prognosis patients were thus treated. The results can be stratified into 3 categories, as follows: I: 12 refractory patients: 3 CR, 6 PR, 2 failures, 1 unevaluable. Median survival was short (7 months, range 2-16 months); II: 6 patients with sensitive relapse: 5 CR, 1 unevaluable. Four patients remained in CR at 42, 45, 46 and 48 months; III: 26 patients who received PEC as consolidation therapy in the first CR-PR after courses of conventional chemotherapy and who were selected at the time of diagnosis on the basis of factors of poor prognosis: 16 CR, 3 PR, 7 unevaluable. Two-year disease-free survival (Kaplan-Meier) is 60%. The median duration of neutropenia (less than 0.5 x 10(9)/l) was 15 d (range 6-37) and that of thrombocytopenia (less than 20 x 10(9)/l), 13 d (range 3-32). The most significant non-hematologic toxicity was of the gastrointestinal tract; in particular severe mucositis. Four iatrogenic deaths occurred (2 candidiasis, 1 aspergillosis and 1 hemorrhage). Pharmacokinetic studies were used to determine the optimum time for reinfusing the marrow and provided information on the mechanism of the mucositis. In conclusion, PEC protocol showed a high efficacy with 94% response rate (68.5% CR) among the 35 evaluable pts. Patients refractory to conventional treatment did not appear to benefit from this intensive chemotherapy. However, it appears that it may be useful for patients with relapses sensitive to salvage therapy. Encouraging results were obtained with the PEC protocol administered as early consolidation for patients with identifiable risk factors at diagnosis. A multicenter, randomized trial is currently underway in France to compare this approach with standard conventional chemotherapy.