Fazarabine has demonstrated a broad spectrum of antitumor activity in experimental models including P388 and L1210 cell lines. Previous phase I clinical trials using a 3-day continuous infusion schedule of Fazarabine have shown myelosuppression to be the dose limiting toxicity in solid tumors. Based on this clinical and preclinical experience we designed a phase I study to determine the toxicity, maximum tolerated dose (MTD), and antileukemic efficacy of Fazarabine using a 3-day continuous infusion schedule in patients with refractory or relapsed acute leukemia or chronic myelogenous (CML) in blastic phase. Adults with a diagnosis of acute leukemia or blastic phase CML who were refractory or had relapsed on salvage chemotherapy were entered on study. Fazarabine was administered as a continuous infusion over 3 days every 3 to 4 weeks. The initial dose was 2 mg/m2/hour x 72 hours. Results showed that the MTD was 425 mg/m2/hour infused over 72 hours every 3 to 4 weeks. At this dose level neurotoxicity and fluid overload were the dose limiting toxicities. Among 71 patients treated, we observed one complete remission, one partial remission and one hematologic improvement. No obvious dose response relationship could be determined. In conclusion, Fazarabine has not shown a beneficial effect in the therapy of acute leukemia. Since 71 patients and 20 dose levels were required to determine the MTD of Fazarabine, a reassessment of our phase I study designs should be considered to provide patients with better potential toxic: therapeutic benefits in such trials.