Eighty patients with type I diabetes and end stage renal disease were prospectively evaluated for coronary artery disease with dipyridamole-thallium-201 scintigraphy and quantitative coronary angiography. Forty patients received dipyridamole orally, and 40 received it intravenously. The prevalence of coronary artery disease was 53%. There were no significant differences in the accuracy of the two dipyridamole tests (sensitivity = 85%, specificity = 85%, accuracy = 85% for the oral group; sensitivity = 86%, specificity = 72%, accuracy = 79% for the intravenous group). Combining the 80 patients into a single group gave a sensitivity of 86%, a specificity of 79%, and an accuracy of 83% for the detection of coronary disease. Although the accuracy of this test in this patient population was similar to that previously reported for other groups, the prevalence of disease was high and resulted in a low predictive value of a negative test (83%).