Etoposide, bleomycin, and cisplatin based combination chemotherapy (BEP) and the surgical removal of residual disease is the standard treatment of metastatic germ cell tumors (GCT). Standard treatment including three cycles of BEP for good risk patients and four cycles of BEP for poor risk patients allows 95 and 65% cure rates in these groups respectively. However, about 10% of patients achieving CR after first line therapy eventually relapse. Conventional second line treatment (VeIP) includes a combination of vinblastin, ifosfamide and cisplatin or (VIP) with etoposide replacing vinblastin for patients pretreated by noncontaining etoposide chemotherapy regimens. These protocols induce a 60% CR rate and a 20-30% long term nonevolutive disease (NED) rate. Trials with high-dose chemotherapy (HDCT) and autologous hematopoietic stem cell transplantation (AHSCT) have been undertaken during the past decade. The main experiences were obtained with combinations of etoposide, platin derivatives, and either cyclophosphamide or ifosfamide. The majority of studies concerned heavily pretreated patients. Few studies have included HDCT and AHSCT as part of consolidation first line treatment for poor risk patients. A randomized multicenter French trial showed no significant benefit of this procedure (Droz, ASCO 1992). In refractory patients, HDCT and AHSCT induces a potential of 10% NED rate and this appears to be the maximum effect that can be obtained. An analysis of prognostic factors at first salvage chemotherapy revealed four adverse factors: high serum levels of HCG (> 10,000 mIU/ml) or AFP (> 1,000 g/ml); extragonadal origin; presence of lung metastases at salvage; and an incomplete response to first line treatment. Three prognostic groups have been defined according to a combination of these variables (Droz, Kramar, ASCO 1993). The use of HDCT followed by AHSCT as consolidation salvage treatment seems more promising since 25-45% of patients have long-term NED. It can be assumed that this procedure may increase the long term NED rate by 20% and produce a success rate up to 40%, but further studies are warranted in this setting. In order to evaluate HDCT with AHSCT rescue, an international randomized study has been undertaken in relapsed or first partial remission GCT patients. Randomization is stratified according to the three prognostic groups previously defined. Salvage regimens for eligible patients include two cycles of PEI or VeIP and in case of response, two more cycles of the same protocol (arm A) or one more cycle of PEI or VeIP followed by HDCT (CarboPEC: carboplatin, etoposide, and cyclophosphamide) (arm B).(ABSTRACT TRUNCATED AT 400 WORDS)