Background-Acetycholine produces coronary artery (CA) constriction in diabetic patients, suggesting an impairment of endothelium-dependent dilation. In diabetes, multiple metabolic abnormalities may inactivate nitric oxide through oxygen free radical production. Methods and Results-To examine the mechanism of this abnormal response, two physiological tests (ie, a cold presser test [CPT] and coronary flow increase induced by an injection of 10 mg papaverine [PAP] in the distal left anterior descending CA) were performed before and after either intravenous L-arginine (625 mg/minx10 minutes) or intravenous deferoxamine (50 mg/minx10 minutes) in 22 normotensive nonsmoking diabetic patients with angiographically normal CAs and normal cholesterol. Coronary surface areas were measured with quantitative angiography. Before the administration of L-arginine or deferoxamine, CPT induced CA constriction ill both groups (-14+/-10% and -15+/-11%, respectively; each P<.001), and PAP injection in distal LAD did not modify significantly proximal LAD dimensions. In the 10 diabetic patients receiving L-arginine, responses to CPT and PAP were not modified. Conversely, in the 12 patients receiving deferoxamine, CA dilated in response to the two tests (+10+/-9% after CPT and +22+/-7% after PAP, each P<.001). Intracoronary isosurbide dinitrate, an endothelium-independent dilator, produced similar dilation in tho two groups (+47+/-19% and +41+/-15%, respectively; each P<.001). Conclusions-This study shows that (1) responses of angiographically normal CAs to CPT and to flow increase are impaired in diabetic patients; (2) abnormal responses are not improved by L-arginine, suggesting that a deficit in substrate fur nitric oxide synthesis is not involved; and (3) deferoxamine restores a vasodilator response to the two tests, suggesting that inactivation of NO by oxygen species might be partly responsible for the impairment of CA dilation in diabetic patients.