Brief episodes of ischemia render the heart more resistant to subsequent ischemia; this phenomenon has been called ischemic preconditioning. In some animal species, myocardial preconditioning appears to be due to activation of ATP-sensitive K+ (KATP) channels. The role played by KATP channels in preconditioning in humans remains unknown. The aim of this study was to establish whether glibenclamide, a selective KATP channel blocker, abolishes the ischemic preconditioning observed in humans during coronary angioplasty following repeated balloon inflations. Twenty consecutive patients undergoing one-vessel coronary angioplasty were randomized to receive 10 mg oral glibenclamide or placebo. Sixty minutes after glibenclamide or placebo administration, patients were given an infusion of 10% dextrose (8 mL/min) to correct glucose plasma levels or, respectively, an infusion of saline at the same infusion rate. Thirty minutes after the beginning of the infusion, both patient groups underwent coronary angioplasty. The mean values (+/- 1 SD) of ST-segment shifts on the surface 12-lead ECG and the intracoronary ECG were measured at the end of the first and second balloon inflations, both 2 minutes long. In glibenclamide-treated patients, the mean ST-segment shift during the second balloon inflation was similar to that observed during the first inflation (23 +/- 13 versus 20 +/- 8 mm, P = NS), and the severity of cardiac pain was greater (55 +/- 21 versus 43 +/- 23 mm on a scale of 0 to 100, P < .05). Conversely, in placebo-treated patients the mean ST-segment shift during the second inflation was less than that during the first inflation (9 +/- 5 versus 23 +/- 13 mm, P < .001), as was the severity of cardiac pain (15 +/- 15 versus 42 +/- 19 mm, P < .01). Blood glucose levels were significantly reduced 60 minutes after glibenclamide compared with those at baseline (53 +/- 9 versus 102 +/- 10 mg/100 mL, P < .001) in the glibenclamide group; however, before coronary angioplasty, blood glucose levels increased to 95 +/- 19 mg/100 mL, a value similar to that found in placebo group (96 +/- 11 mg/100 mL, P = NS). In humans, ischemic preconditioning during brief repeated coronary occlusions is completely abolished by pretreatment with glibenclamide, thus suggesting that it is mainly mediated by KATP channels.