Volatile anesthetics stimulate, but hyperglycemia attenuates, the activity of mitochondrial ATP-regulated K(+) channels. We tested the hypothesis that diabetes mellitus interferes with isoflurane-induced preconditioning. Acutely instrumented, barbiturate-anesthetized dogs were randomly assigned to receive 0, 0.32, or 0.64% end-tidal concentrations of isoflurane in the absence or presence of diabetes (3 wk after administration of alloxan and streptozotocin) in six experimental groups. All dogs were subjected to a 60-min left anterior descending coronary artery occlusion followed by 3 h of reperfusion. Myocardial infarct size (triphenyltetrazolium staining) was 29 +/- 3% (n = 8) of the left ventricular area at risk in control experiments. Isoflurane reduced infarct size (15 +/- 2 and 13 +/- 1% during 0.32 and 0.64% concentrations; n = 8 and 7 dogs, respectively). Diabetes alone did not alter infarct size (30 +/- 3%; n = 8) but blocked the protective effects of 0.32% (27 +/- 2%; n = 7) and not 0.64% isoflurane (18 +/- 3%; n = 7). Infarct size was directly related to blood glucose concentrations in diabetic dogs, but this relationship was abolished by higher concentrations of isoflurane. The results indicate that blood glucose and end-tidal isoflurane concentrations are important determinants of infarct size during anesthetic-induced preconditioning.