Background: Patients referred for stress echocardiography may differ significantly from those referred for stress myocardial perfusion imaging (MPI). Such differences, if present, should be considered when comparing the accuracy and discriminatory performance of these tests. Methods: We prospectively collected demographic and clinical information on all stress imaging studies performed at our institution between 1998 and 2001. The data were reviewed, summarized, and compared using the t test and χ2 test where appropriate. Results: Of 5320 stress imaging studies performed, 3383 were stress echocardiographies and 1937 were MPI studies. Patients referred for MPI were older (59 vs 54, P < .0001), and more likely to have diabetes (32% vs 20%, P < .0001), prior myocardial infarction (39% vs 15%, P < .0001), and prior revascularization (38% vs 12%, P < .0001). Pharmacologic stress testing was much more common in the MPI group (66% vs 17%, P < .0001). More patients referred for MPI had decreased left ventricular function (23% vs 7%, P < .0001) and abnormal stress test results (41% vs 18%, P < .0001). Conclusions: Patients with a history of myocardial infarction, revascularization, or higher risk profiles are more likely to be referred for MPI compared to stress echocardiography at our institution. These differences in referral patterns are likely to exist in other centers, and it is reasonable to assume that systematic differences in test selection occur, resulting in patient populations with differing clinical risk profiles. Caution in interpreting analyses comparing the accuracy of stress imaging modalities is appropriate.