Complete remission and long-term survival rates are low for older adults treated for acute myeloid leukemia (AML). Because of favorable phase 2 data using mitoxantrone and etoposide, we conducted a phase 3 study (SWOG-9333) in which patients over 55 years of age with previously untreated AML were randomized to receive mitoxantrone (10 mg/m(2) per day x 5) and etoposide (100 mg/m(2) per day x 5) [ME], or cytarabine (200 mg/m(2) per day x 7) and daunorubicin (45 mg/m(2) per day x 3) [AD] as induction therapy. The randomization was stratified by age, onset of leukemia, and multidrug resistance phenotype. Over a 4-year period, 328 eligible patients from 66 institutions were enrolled. The complete remission rate was 34% (95% confidence interval [CI] 26%-41%) for patients in the ME and 43% (CI 35%-51%) for patients in the AD treatment arm (one-tailed P value.96). The rates of resistant disease were 43% (CI 35%-51%) and 34% (CI 27%-42%), respectively, for the 2 treatment arms (one-tailed P value.95). The estimated overall survival at 2 years was 11% (CI 6%-15%) and 19% (CI 12%-25%) for patients randomized to ME and to AD induction therapy, respectively (one-tailed P value.99). After accounting for the independent prognostic factors associated with survival (karyotype, performance status, age, white blood cell count), exploratory analysis suggested there was a worse survival for patients who received ME compared with AD induction therapy (2-tailed P value.0066). We conclude that the results of our study do not demonstrate any benefit to the use of ME induction chemotherapy instead of AD in older patients with AML.