Interest in high-dose cytarabine (HDAC) for both induction and postremission therapy for acute myeloid leukemia (AML) prompted the Southwest Oncology Group (SWOG) to initiate a randomized trial comparing HDAC with standard-dose cytarabine (SDAC) for remission induction of previously untreated AML and to compare high-dose treatment versus conventional doses for consolidation therapy. Patients less than 65 years of age with de novo or secondary AML were randomized for induction between SDAC 200 mg/ m2/d for 7 days by continuous infusion or HDAC at 2 g/ m2 intravenously every 12 hours for 12 doses; both groups received daunorubicin (DNR) at 45 mg/m2/d intravenously for 3 days. Complete responders to SDAC were randomized to receive either two additional courses of SDAC plus DNR or one course of HDAC plus DNR. Complete responders to HDAC were nonrandomly assigned to receive one additional course of HDAC plus DNR. Of patients randomized between SDAC (n = 493) and HDAC (n = 172) induction, 361 achieved complete remission (CR). The CR rate was slightly poorer with HDAC: 55% versus 58% with SDAC for patients aged less than 50, and 45% (HDAC) versus 53% (SDAC) for patients aged 50 to 64 (age-adjusted one-tailed P = .96). With a median follow-up time of 51 months, survival was not significantly better with HDAC (P = .41); the estimated survival rate at 4 years was 32% (HDAC) versus 22% (SDAC) for those aged less than 50, and 13% (HDAC) versus 11% (SDAC) for those aged 50 to 64. However, relapse-free survival was somewhat better following HDAC Induction (P = .049): 33% (HDAC) versus 21% (SDAC) at 4 years for those aged less than 50, and 21% (HDAC) versus 9% (SDAC) for those aged 50 to 64. Induction with HDAC was associated with a significantly increased risk of fatal (P = .0033) and neurologic (P < .0001) toxicity. Among patients who achieved CR with SDAC, survival and disease-free survival (DFS) following consolidation randomization were not significantly better with HDAC compared with SDAC (P = .77 and .46, respectively). Patients who received both HDAC induction and consolidation had the best postremission outcomes; however, the proportion of CR patients who did not go on to protocol consolidation therapy was more than twice as high after HDAC induction compared with SDAC. Induction therapy with HDAC plus DNR was associated with greater toxicity than SDAC plus DNR, but with no improvement in CR rate or survival. Following CR induction with SDAC, consolidation with HDAC increased toxicity but not survival or DFS. In a nonrandomized comparison, patients who received both HDAC induction and consolidation had superior survival and DFS compared with those who received SDAC induction with either SDAC or HDAC consolidation.