Seventeen coronary patients (CAD) underwent thallium (TI-201) treadmill and radionuclide (RNV) ejection fraction supine bicycle testing before and after 5.6 ± 1.6 ( X ± SD) months of an exercise program. Thallium data were assessed both using analog images and a computerized circumferential profile technique. Patients exercised on the treadmill to a higher workload after the exercise program, but achieved a similar pressure-rate product. When interpreting the analog thallium images, only 50% agreement was obtained for the assessment of changes in myocardial perfusion (pre/post-training). The computer technique, however, had low inter-intraobserver variability (6%) and better agreement (90.5%). Using the circumferential profile method, five patients improved (a total of 11 regions) and one patient worsened (with two regions). Before the exercise program, the ejection fraction (EF) response to supine bike exercise was normal (an increase greater than 11%) in four, flat in seven, and severely abnormal (a decrease of more than 4%) in six patients. After the exercise program, even though achieving similar or higher pressure-rate products, six patients improved their EF response, nine did not change, and two worsened. Of the five patients who improved their thallium images, one improved his EF response, two remained normal, and two did not change. One patient worsened both his thallium study and the EF response after the exercise program. Changes in thallium exercise images and the EF response to supine exercise occurred in our patients after an exercise program, but were not always concordant. Indeed, of five patients with exercise-induced ischemic ST changes before and after training, the EF response improved in three whereas myocardial perfusion was unchanged. Reasons for this lack of agreement are discussed, and have been considered in the planning of a randomized trial of the effects of an exercise program on myocardial perfusion and function.