Chemomodulation of cytarabine by fludarabine has been attributed with a higher antileukemic efficacy, but randomized trials to address this question are rare. We therefore conducted a multicenter, randomized phase III study to evaluate the antileukemic efficacy of adding fludarabine to sequential high-dose cytarabine+idarubicin (SHAI) re-induction chemotherapy in relapsed or refractory acute myeloid leukemia (AML). Patients (n=326, of which 281 were evaluable) were randomly assigned to SHAI (cytarabine, 1 g/m(2) bid, days 1-2 and 8-9 (3 g/m(2) for patients ≤ 60 years with refractory AML or ≥ 2nd relapse); idarubicin 10 mg/m(2) daily, days 3-4 and 10-11) or F-SHAI (SHAI with fludarabine, 15 mg/m(2), 4 h before cytarabine). Although complete remission (CR) rates (35% SHAI and 44% F-SHAI) and overall survival did not differ between both regimens, fludarabine prolonged time to treatment failure from 2.04 to 3.38 months (median, P<0.05). Twenty-seven percent of patients proceeded to allogeneic stem cell transplantation, with a significantly higher number of patients in CR or incomplete remission in the F-SHAI group (22 vs 10%, P<0.01). In conclusion, fludarabine has a beneficial, although moderate, impact on the antileukemic efficacy of high-dose cytarabine-based salvage therapy for relapsed and refractory AML.