The value of exercise positron emission computed tomography (PET) was determined by comparing it with exercise T1-201 single photon emission computed tomography (SPECT) using graded ergometer exercise in 10 anginal patients with left anterior descending coronary (LAD) involvement (stenosis greater than or equal to 75%) with or without other coronary stenosis. The distribution of N-13 ammonia or T1-201 was determined and expressed as percent count ot the highest count on the circumferential profile curve (CPC) in the transaxial image. Two standard deviations below the mean counts of six normal hearts for PET and ten for SPECT were considered as the lower limit of normal, and these were determined separately during exercise and at rest. The ischemic areas per total myocardial areas (%C), the accuracy in diagnosing coronary involvement and the identification of coronary branch involvement were determined by CPC analysis in three corresponding slices. Regional analyses revealed that accuracy in diagnosing proximal left anterior descending artery (LAD) involvement was higher with PET (9/10) than with SPECT (7/10), but the difference was not statistically significant. The accuracy in diagnosing right coronary or circumflex artery stenosis was 9/10 with PET and 8/10 with SPECT (NS); the %C was significantly higher with PET than with SPECT during exercise (PET: mean +/- SD = 56 +/- 21%, SPECT: 35 +/- 21%, p less than .01) in spite of smaller double products (x10(3] (PET: 18.9 +/- 4.7, SPECT: 23.5 +/- 5.7, p less than .02). Thus, exercise PET was more reliable in evaluating regional ischemia and in semiquantifying the ischemic area in the myocardium.