Our previous phase II trial for treating human T-lymphotropic virus type I-associated adult T-cell leukemia-lymphoma (ATLL) with vincristine, cyclophosphamide, doxorubicin, and prednisone (VCAP), doxorubicin, ranimustine, and prednisone (AMP), and vindesine, etoposide, carboplatin, and prednisone (VECP) showed promising results. To test the superiority of VCAP-AMP-VECP over biweekly cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP), we conducted a randomized controlled trial exclusively for ATLL. Previously untreated patients with aggressive ATLL were assigned to receive either six courses of VCAP-AMP-VECP every 4 weeks or eight courses of biweekly CHOP. Both treatments were supported with granulocyte colony-stimulating factor and intrathecal prophylaxis. A total of 118 patients were enrolled. The complete response (CR) rate was higher in the VCAP-AMP-VECP arm than in biweekly CHOP arm (40% v 25%, respectively; P = .020). Progression-free survival rate at 1 year was 28% in the VCAP-AMP-VECP arm compared with 16% in the CHOP arm (P = .100, two-sided P = .200). Overall survival (OS) at 3 years was 24% in the VCAP-AMP-VECP arm and 13% in the CHOP arm (P = .085, two-sided P = .169). For VCAP-AMP-VECP versus biweekly CHOP, grade 4 neutropenia, grade 4 thrombocytopenia, and grade 3 or 4 infection rates were 98% v 83%, 74% v 17%, and 32% v 15%, respectively. There were three toxic deaths in the VCAP-AMP-VECP arm. The longer OS at 3 years and higher CR rate with VCAP-AMP-VECP compared with biweekly CHOP suggest that VCAP-AMP-VECP might be a more effective regimen at the expense of higher toxicities, providing the basis for future investigations in the treatment of ATLL.