One hundred and forty-five adults with acute leukemia were randomized to receive remission induction therapy in or out of a protected environment (PE) with prophylactic antibiotics orally (PA) or systemically (SA). Sixty-three patients were randomized in PE and 82 outside a PE. The proportion of patients who survived long enough to receive an adequate trial was higher in the PE (97%) than out (82%) (P = .01). The complete remission (CR) rate was 71% in and 43% out of the PE (P less than .01). Fifty-five patients received PA and 90 received SA. The CR rates were 61% and 45%, respectively. Of the 145 patients, 73 (50%) developed 102 episodes of major infections. Twenty-six of 63 patients in the PE developed major infection compared to 47 of 82 outside a PE (P = .08). The incidence rate of 13% fatal infections in a PE was significantly smaller than the 28% rate outside a PE (P = .04). The number of days with infections at less than 500 neutrophils/mm3 was also significantly lower inside a PE than outside (P less than .01). When comparing patients receiving SA or PA, there was no statistically significant difference in the incidence of infections. Forty-one patients received OAP Chemotherapy and 104 received adriamycin-OAP plus BCG. The CR rate on OAP was 44% compared with 60% on Ad-OAP + BCG. Infection rates were 76% and 40%, respectively (P less than .01). The median survival time was 72 weeks for patients in PE compared with 42 weeks for patients outside a PE (P less than .01). The prophylactic antibiotic regimens were well tolerated by most patients. This prospective randomized study has demonstrated statistically significant advantages for a lowered risk of fatal infection, higher CR rate and longer survival of patients treated in a PE with prophylactic antibiotics compared with patients treated in a conventional hospital room. Also, there was evidence for the superiority of adriamycin-OAP + BCG treatment compared with OAP.