Thallium tomographic imaging and exercise electrocardiography were performed on 136 diabetic patients without symptoms of heart disease. Thirty three patients had post-exercise thallium defects and 19 had ST 1 mm greater than or equal to segment depression during exercise electrocardiography. Both tests were positive in 13 patients. Coronary angiography was subsequently performed on 33 patients with either scintigraphic and/or electrocardiographic evidence of myocardial ischaemia. Angiographically significant coronary artery disease (greater than or equal to 50% narrowing of the coronary artery lumen) was detected in 13 patients. Six patients had minimal coronary artery stenosis (less than 50%), and 14 had normal coronary arteries. Six patients refused cardiac catheterisation. In 14 out of 27 patients with post-exercise thallium defects coronary angiography did not show any coronary artery stenoses (positive predictive accuracy 48%). Exercise electrocardiography showed only one false positive result (positive predictive accuracy 94%) but failed to detect coronary artery disease in three patients with a positive scintigraphic result. The accuracy of a positive exercise electrocardiographic test seems to be better than that of a positive thallium tomographic scan for detecting asymptomatic coronary artery disease in diabetic patients. The high number of false positive thallium defects may be the result of technical features inherent in thallium tomography and/or the possible disease of the small intramyocardial arteries in diabetic patients.