OBJECTIVE: The aim was to examine whether diabetes interferes with the development of myocardial injury in a canine ischaemia-reperfusion model. METHODS: Non-insulin-requiring diabetes was induced in dogs by the streptozotocin-alloxan method. After 75 d, the dogs were anaesthetised and myocardial infarction was provoked by occluding the left anterior descending coronary artery for 2 h followed by 6 h reperfusion. RESULTS: Diabetic dogs had higher blood glucose [9.4(SEM 1) mmol.litre-1], fructosamine [417(57) mumol.litre-1], and glycated haemoglobin [3.3(0.7)%], than control dogs [5.5(0.6), p = 0.04, 243(15), p = 0.01, and 0.7(0.2), p = 0.003, respectively], and they also had higher serum lipids (p = 0.001) and platelet aggregation (p = 0.03). Area at risk was similar in diabetic and control dogs but in contrast to controls (r = 0.78, p = 0.007), area at risk and infarct size were not correlated in diabetics (r = 0.08). In both groups, collateral flow was the major determinant of infarct size: r = -0.73 in controls (p = 0.02) and -0.97 in diabetics (p = 0.001). In spite of higher subendocardial collateral flow in diabetics [representing 21.6(6)% of the flow in the corresponding non-ischaemic zone] than in controls [11.2(6)%], infarct size was similar in both groups. However, the mean observed infarct size in the diabetic group [7.5(2.8)% of the left ventricle] was significantly (p < 0.03) larger than the mean predicted infarct size [5.2(2)%]. Multivariate analysis confirmed that diabetes, as well as collateral flow, is an independent (p = 0.03) predictor of infarct size. CONCLUSIONS: For a given collateral flow, diabetic dogs develop larger infarcts than controls. Further studies are required to investigate the biochemical mechanism(s) underlying this deleterious effect. However, this may partly explain the poor prognosis of myocardial infarction in diabetic persons.