This study assesses the influence of left ventricular hypertrophy (LVH) on the accuracy of myocardial perfusion imaging using pharmacologic coronary vasodilation. Seventy-five patients without previous infarction, and with known coronary anatomy, were studied by echocardiography and PET. LVH (defined by mass greater than 131 g/m2 in males or greater than 100 g/m2 in females) was identified in 25 patients; this group did not differ significantly from the remainder in terms of clinical or angiographic parameters. Twenty patients with hypertrophy had significant coronary artery stenoses, which were identified correctly by PET in 11 (55%), in contrast to 29 of 34 patients (85%, p = 0.03) with coronary disease but normal LV mass. Normal perfusion images were obtained in three of five patients (60%) with hypertrophy but no coronary disease; in contrast, 14 of 16 patients without either coronary disease or hypertrophy (88%, p = ns) had normal scans. The accuracy of PET was 14/25 (56%) in those with hypertrophy, and 43/50 (86%, p = 0.01) in patients with normal LV mass. In this group, the presence of hypertrophy was associated with reduction in the diagnostic accuracy of PET using dipyridamole stress. These findings may account for the phenomenon of "dipyridamole nonresponsiveness" in some patients.