Between 1982 and 1986, 326 evaluable patients with acute myeloid leukemia (AML) were randomized to receive cytarabine (Ara-C) at 200 mg/m2 (A200) or 100 mg/m2 (A100) for induction and maintenance therapy. Cycle 1 of induction therapy consisted of 7 days of continuous intravenous (IV) Ara-C and 3 days of i.v. daunorubicin (DNR); cycle 2, if needed, consisted of 5 days of Ara-C and 2 days of DNR. Complete responders (CR) then received monthly subcutaneous (SC) Ara-C at the respective doses (A100 or A200) with 6-thioquanine (6TG) at months 1 and 5, with vincristine (VCR) and prednisone at months 2, 4, 6, and 8, and with DNR at months 3 and 7. Complete response rates were 58% (A100) and 64% (A200) (P = .29). Median survival was 46 weeks (A100) and 38 weeks (A200) (P = .64); 5-year survival was 10% (A200) and 8% (A100). Median time to remission was 6.7 weeks (A200) and 8.1 weeks (A100) (P = .18). Median disease-free survival was 41 weeks (A200) and 44 weeks (A100) (P = .86). Deaths were attributed to therapy-related toxicities in 21% (A200) and 13% (A100) (P = .05). The 5-year survival was 15% for patients with performance status (PS) 0, 8% for PS 1 to 2, and 2% for PS 3 to 4, 18% for patients less than 40 years, 8% for ages 40 to 59, and 3% for age 60 or greater. Stratification of data by age and PS suggested that A200 may improve survival in patients less than 60 years with a good PS 0 (P = .05). This trial does not support the superiority of A200 over A100 in the treatment of AML.