BACKGROUND Cisplatin-based combination chemotherapy will cure 70% to 80% of patients with metastatic non-seminomatous germ cell tumors but is associated with the possibility of severe neuro-, oto- and nephro-toxicities. Carboplatin, a cisplatin analogue, is an active drug in testicular cancer with a more favourable spectrum of side effects. In a randomized trial, the German Testicular Cancer Study Group compared a combination regimen of carboplatin, etoposide and bleomycin (CEB) to standard cisplatin, etoposide and bleomycin (PEB) chemotherapy for patients with 'minimal-' and moderate-disease' non-seminomatous germ cell tumors, according to the Indiana University classification. METHODS PEB was given for three cycles at standard doses (given days 1-5), and the CEB regimen consisted of carboplatin (target AUC of 5 mg/ml x min) on day 1, etoposide 120 mg/m2 on days 1 to 3 and bleomycin 30 mg on days 1, 8 and 15. Four cycles of CEB were given, with the omission of bleomycin in the fourth cycle. Thus, the cumulative doses of etoposide and bleomycin applied in the two treatment arms were comparable. Fifty-four patients were entered on the trial, 29 were treated with PEB and 25 with CEB chemotherapy. Patients were stratified according to disease extent (minimal versus moderate) and the degree of tumor marker elevation. Thirty-two patients (59%) belonged to the group with minimal disease and low markers. RESULTS No significant difference in response to chemotherapy was seen between the two arms, with CR rates of 81% for the PEB arm and 76% for CEB treatment. However, more patients treated with CEB (32% versus 13%) have relapsed after therapy, and 4 patients (16%) have died of disease progression after CEP in contrast to 1 (3%) after PEB therapy. The first interim analysis of negative events (relapse, vital tumor at secondary resection, death from disease and therapy-associated death) showed a significantly higher rate after CEB than after PEB therapy, and the trial was terminated early. After a median follow-up of 33 months for all patients, the calculation of negative events is still significantly in favour of PEB-treated patient, particularly since three late relapses > 2 years have been observed in the CEB arm (P = 0.03). CONCLUSIONS This randomized trial demonstrates that even with the use of adequate doses of etoposide and full-dose bleomycin, carboplatin cannot altogether replace cisplatin in patients with testicular cancer. Treatment with the PEB regimen remains the standard approach in patients with 'good-risk' non-seminomatous germ cell tumors.