Advances in the treatment of acute myeloid leukemia (AML) have occurred with the introduction of new therapies including high-dose cytarabine and the identification of powerful prognostic factors such as cytogenetics that predict for long-term outcome. To date, the prognostic impact of cytarabine dose escalation within various cytogenetic groups of AML has not been assessed. We describe 285 newly diagnosed patients with primary AML who had adequate karyotypes and were enrolled on a prospective Cancer and Leukemia Group B cytogenetic study. All patients were randomly assigned to postremission treatment with standard-, intermediate-, or high-dose cytarabine intensification. Patients were categorized to one of three cytogenetic groups: (a) core binding factor type [(CBF); i.e., t(8;21) inv(16), t(16;16), and del(16)]; (b) normal; and (c) other abnormality karyotype. An evaluation of these patients after a median follow-up time of over 7 years was performed to determine the relationship of intensification to outcome by cytogenetic group. Patients included 57 patients with CBF AML, 140 patients with normal karyotype AML, and 88 patients with other cytogenetic abnormalities. The treatment outcome of CBF AML patients was superior, with an estimated 50% still in complete remission (CR) after 5 years as compared with 32 and 15% for patients with normal karyotype AML and other abnormality AML, respectively (P < 0.001). Univariate analysis showed the following nonkaryotype factors to predict a prolonged CR duration: (a) younger age (P < 0.008); (b) lower leukocyte count (P = 0.01); (c) the presence of Auer rods (P = 0.004); (d) a lower percentage of bone marrow blasts (P = 0.001) at the time of diagnosis, (e) and a higher postremission cytarabine dose (P < 0.001). The impact of cytarabine dose on long-term remission was most marked (P < 0.001) in the CBF AML group (after 5 years, 78% of those with a dose of 3 g/m2 were still in CR, 57% of those with a dose of 400 mg/m2 were still in CR, and 16% of those with a dose of 100 mg/m2 were still in CR) followed by normal karyotype AML (P = 0.01; after 5 years, 40% of those with a dose of 3 g/m2 were still in CR, 37% of those with a dose of 400 mg/m2 were still in CR, and 20% of those with a dose of 100 mg/m2 were still in CR). In contrast, cytarabine at all doses produced only a 21% or less chance of long-term continuous CR for patients with other cytogenetic abnormalities. A multivariate analysis of CR duration assessed the independent impact of each of these variables on cure. Significant factors entering this model in descending order of importance were cytogenetic group (CBF > normal > other abnormality; P = 0.00001), cytarabine dose (3 g/m2 > 400 mg/m2 > 100 mg/m2; P = 0.00001), logarithm of leukocyte count at the time of diagnosis (P = 0.0005), and histological subtype of AML (P = 0.005). This study demonstrates that the curative impact of cytarabine intensification varies significantly among cytogenetic groups and results in a substantial prolongation of CR among patients with CBF and normal karyotypes, but not in those with other karyotypic abnormalities. These findings support the use of pretreatment cytogenetics in risk stratification of postremission AML therapy.