A phase III prospective randomized multicenter study was performed to determine whether quinine could improve the response rate of poor-risk acute leukemias (ALs) to standard chemotherapy including a multidrug resistance (MDR)-related cytotoxic agent. The rationale of the study was based on the negative prognostic value of MDR phenotype in ALs and the ability of quinine to reverse this phenotype both in vitro and ex vivo. Three hundred fifteen patients (median age, 49 years; range, 16 to 65) with relapsed (n = 108) or refractory (n = 32) acute myeloblastic leukemia (AML), relapsed (n = 27) or refractory (n = 9) acute lymphoblastic leukemia (ALL), secondary AL (n = 22) or blastic transformation of myelodysplastic syndrome ([MDS] n = 74) or myeloproliferative syndrome ([MPS] n = 43) were randomly assigned to receive mitoxantrone ([MXN] 12 mg/m2/d, days 2 to 5) and cytarabine ([Ara-C] 1 g/m2/12 h, days 1 to 5) alone or in combination with quinine (30 mg/kg/d, days 1 to 5; continuous intravenous infusion beginning 24 hours before MXN infusion). Side effects of quinine were observed in 56 of 161 quinine-treated patients and disappeared in all but four cases after one or two 20% dose decreases. Sera from quinine-treated patients showed increased MXN uptake in an MDR-positive cell line compared with matched sera obtained before quinine infusion. Quinine induced a significant increase in the incidence of nausea, vomiting, mucositis, and cardiac toxicity. A complete response (CR) was observed in 85 of 161 patients (52.8%) from the quinine-treated group versus 70 of 154 patients (45.5%) in the control group (P = .19). The most important differences between quinine and control group CR rates were observed in patients with refractory AMLs and blastic transformation of MDS and MPS. The CR rate was higher in P-glycoprotein-positive cases, although the difference was not significant. Failure of the regimen due to blastic persistence or blast number increase was higher in the control group (61 of 154 patients) than in the quinine group (45 of 161, P = .04). Early death was observed in eight cases (four in each arm) and death in aplasia in 27 cases (20 in quinine group v seven in control group, P = .01). The significant increase of toxicity in the quinine arm could have masked the clinical benefit of MDR reversion in poor-risk ALs.