Purpose Extramedullary disease (EMD) at diagnosis in patients with acute myeloid leukemia (AML) has been recognized for decades. Reported herein are results from a large study of patients with AML who were treated in consecutive ECOG-ACRIN Cancer Research Group frontline clinical trials in an attempt to define the incidence and clinical implications of EMD. Methods Patients with newly diagnosed AML, age 15 years and older, who were treated in 11 clinical trials, were studied to identify EMD, as defined by physical examination, laboratory findings, and imaging results. Results Of the 3,522 patients enrolled, 282 were excluded, including patients with acute promyelocytic leukemia, incorrect diagnosis, or no adequate assessment of EMD at baseline. The overall incidence of EMD was 23.7%. The sites involved were: lymph nodes (11.5%), spleen (7.3%), liver (5.3%), skin (4.5%), gingiva (4.4%), and CNS (1.1%). Most patients (65.3%) had only one site of EMD, 20.9% had two sites, 9.5% had three sites, and 3.4% had four sites. The median overall survival was 1.035 years. In univariable analysis, the presence of any EMD ( P = .005), skin involvement ( P = .002), spleen ( P < .001), and liver ( P < .001), but not CNS ( P = .34), nodal involvement ( P = .94), and gingival hypertrophy ( P = .24), was associated with a shorter overall survival. In contrast, in multivariable analysis, adjusted for known prognostic factors such as cytogenetic risk and WBC count, neither the presence of EMD nor the number of specific sites of EMD were independently prognostic. Conclusion This large study demonstrates that EMD at any site is common but is not an independent prognostic factor. Treatment decisions for patients with EMD should be made on the basis of recognized AML prognostic factors, irrespective of the presence of EMD.