Patients with acute nonlymphocytic leukemia were randomized to receive remission induction therapy consisting of seven days of cytosine arabinoside and three days of daunorubicin ("7 + 3") or to receive the same regimen intensified by either the addition of 6-thioguanine or by extension of the administration of cytosine arabinoside to ten days. Additionally, all patients were randomized to receive or not to receive cotrimoxazole antibacterial prophylaxis during the remission induction phase. Neither an increase in intensity of chemotherapy nor the antibacterial prophylaxis increased the remission rate above the 53% for patients treated with the standard "7 + 3" regimen. The second part of this study addressed the issue of the utility of long-term maintenance chemotherapy. To this end, patients were randomized to discontinue all treatment after 8 months of maintenance chemotherapy or to continue maintenance therapy for a total of 3 years. Although there was a transient increase in the relapse rate for patients who discontinued therapy, the proportion of long-term remitters was identical in the two patient groups. Additionally, there is a suggestion of a survival advantage for patients randomized to discontinue all therapy at 8 months.