We performed a prospective randomized trial of a high-dose chemotherapy regimen v standard cisplatin-based chemotherapy in poor prognosis nonseminomatous germ-cell cancer patients. The high-dose regimen consisting of twice the standard dose of cisplatin (P), along with vinblastine (Ve), bleomycin (B), and the epipodophylotoxin etoposide (VP-16) (V) (PVeBV) was compared to the classic regimen with normal dose cisplatin, vinblastine, and bleomycin (PVeB). Eligibility criteria included large abdominal masses, liver metastases, multiple pulmonary metastases, brain metastases, marked elevations in serum tumor markers (alpha-fetoprotein greater than 1,000 ng/mL or the beta-subunit of human chorionic gonadotropin greater than 10,000 mIU), unfavorable histology (pure choriocarcinoma), or extragonadal germ-cell tumors. Fifty-two consecutive patients with poor prognostic features were randomized to receive either PVeBV or PVeB. The median follow-up is 4 years. Treatment with the high-dose regimen increased the complete remission rate (88% v 67%, P = .14) and was associated with a lower relapse rate (17% v 41%, P = .2). The median survival of patients receiving standard therapy was 30 months, while the median survival for patients receiving the high-dose regimen has not been reached. Actuarial 5-year survival for patients treated with the high-dose regimen is 78%, compared with 48% for patients receiving standard therapy (two-sided Mantel-Cox test = .06). Disease-free survival was also superior for patients randomized to PVeBV (P = .03). Sixty-eight percent of patients (23 of 34) randomized to PVeBV are alive and continuously disease-free, compared with 33% (six of 18) for PVeB (P = .02). The major difference in toxicity between the high-dose regimen and standard therapy was the severity of myelosuppression and the incidence of severe hearing loss. Ninety-one percent of patients treated with PVeBV had a WBC count less than 1,000/microL, compared with 50% of patients receiving PVeB (P less than .05). Hearing aids were recommended for 12 patients who received PVeBV and two who received PVeB. The increased effectiveness of the PVeBV regimen in poor prognosis germ-cell cancer patients may relate to the double-dose cisplatin, the addition of VP-16, or to a synergistic effect of these two drugs.