This study compared the accuracy of cardiovascular diagnoses by medical students operating a small hand-carried ultrasound (HCU) device with that of board-certified cardiologists using standard physical examinations. Sixty-one patients (38% women; mean age 70 +/- 19 years) with clinically significant cardiac disease had HCU studies performed by 1 of 2 medical students with 18 hours of training in cardiac ultrasound and physical examinations by 1 of 5 cardiologists. Diagnostic accuracy was determined by standard echocardiography. Two-hundred thirty-nine abnormal findings were detected by standard echocardiography. The students correctly identified 75% (180 of 239) of the pathologies, whereas cardiologists found 49% (116 of 239) (p <0.001). The students' diagnostic specificity of 87% was also greater than cardiologists' specificity of 76% (p <0.001). For nonvalvular pathologies (115 findings), students' sensitivity was 61%, compared with 47% for cardiologists (p = 0.040). There were 124 clinically significant valvular lesions (111 regurgitations, 13 stenoses). Students' and cardiologists' sensitivities for recognizing lesions that cause a systolic murmur were 93% and 62% (p <0.001), respectively. Students' sensitivity for diagnosing lesions that produce a diastolic murmur was 75%; cardiologists recognized 16% of these lesions (p <0.001). The diagnostic accuracy of medical students using an HCU device after brief echocardiographic training to detect valvular disease, left ventricular dysfunction, enlargement, and hypertrophy was superior to that of experienced cardiologists performing cardiac physical examinations.