Current intensive chemotherapy for acute nonlymphoblastic leukemia (ANLL) results in a complete remission in the majority of patients. Unfortunately, the duration of remission is short and most of the patients will experience a relapse of their underlying disease. Autologous bone marrow (BM) transplantation is being explored as a treatment modality designed to improve relapse-free survival. We have conducted a phase II trial exploring the combination of busulfan (16 mg/kg) and etoposide (60 mg/kg) in an attempt to improve antitumor efficacy using this novel preparative regimen. To date, 50 patients (48 with ANLL and 2 patients with biphenotypic acute leukemia) have been treated. The first 20 patients received unmanipulated BM; 28 patients subsequently received 4-hydroperoxycyclophosphamide (4-HC) (60 micrograms/mL)-purged bone marrow, and 2 patients with biphenotypic acute leukemia received both 4-HC (60 micrograms/mL) and etoposide (5 micrograms/mL)-purged BM. Thirty-four patients were in first complete remission (CR1), 12 patients in second complete remission (CR2), and 4 patients in relapse. The median time from first complete remission to BM harvest was 3 months (range, 0.8 to 4) compared with median time of 2 months (range, 1.5 to 5.0) for patients in second complete remission. The median time from harvest to transplant was 1 month for both groups (range, 0.4 to 36). A median of 0.7 x 10(8) (range, 0.2 to 1.4) mononuclear cells were infused. Patients achieved an absolute neutrophil count of > or = 500/microL at a median of 26 days (range, 13 to 96), an untransfused platelet count > or = 20,000/microL at a median of 56 days (range, 15 to 278) and a sustained hematocrit > or = 30% at a median of 50 days (range, 19 to 116). Twenty-six patients are alive and in continued CR. Follow-up of the surviving patients ranged from 6 months to 66 months with a median follow-up of 31 months. Patients receiving purged BM have an actuarial disease-free survival of 57% with a relapse rate of 28% compared with patients receiving unpurged BM whose actuarial disease-free survival is 32% with a relapse rate of 62% (P = .06 for relapse rate). The most significant extramedullary toxicities for this regimen are hepatic and cutaneous (including mucositis). The BU/VP-16 regimen is associated with a significant proportion of patients surviving disease free, especially in the group receiving purged BM. Whether this regimen offers a substantial improvement in disease-free survival over currently used regimens will require a prospective randomized study.