OBJECTIVE To compare the prevalence of silent myocardial ischaemia associated with coronary artery disease in diabetic subjects with that in controls of similar age and sex. METHODS A controlled study in which subjects with positive findings on exercise electrocardiography, 24 hour electrocardiographic recording, or dynamic thallium scintigraphy (diabetics only) underwent coronary angiography. METHODS Academic medical centre; referral based cardiology clinic. METHODS 136 Diabetic subjects, of whom 72 (33 women, 39 men (mean age 46.0] were insulin dependent and 64 (19 women, 45 men (mean age 49.3] non-insulin dependent. 80 Controls matched for age and sex; all were clients of the Occupational Health Service of Oulu University Central Hospital or the State Occupational Health Service Station in Oulu in whom diabetes had been excluded by a glucose tolerance test. METHODS Any subject showing signs of myocardial ischaemia was referred for cardiac catheterization. METHODS Exercise electrocardiography and 24 hour electrocardiographic recording were regarded as positive if there were ST depressions of greater than or equal to 1 mm that were planar or downsloping and persisted for 0.08 seconds after the J point. Thallium tomographic imaging. With cardiac catheterisation, coronary artery lesions were classified as significant in half or more of the vessel lumen was narrowed, or insignificant if such narrowing was less than half. RESULTS 40 (29%) diabetes and four (5%) controls had positive results in one or more of the non-invasive tests. Coronary angiography was performed on 34 of the diabetics (six refused); 12 had significant coronary artery narrowing; seven had unimportant atherosclerosis; 15 had patent coronary arteries. Among the controls only one had unimportant atherosclerosis; the other three had patent arteries. CONCLUSIONS These results confirm the high prevalence of asymptomatic myocardial ischaemia in diabetics. Non-invasive screening of diabetic subjects, however, does not seem justified because of the low preset probability of the presence of the disease and the inaccuracy of the available test methods.