1. We measured endothelium-dependent and independent dilatation of forearm resistance arteries in 29 men with diet-treated non-insulin-dependent diabetes mellitus and 18 age- and sex-matched control subjects. None of the diabetic patients had hypercholesterolaemia, overt hypertension or microproteinuria. 2. We examined endogenous and exogenous nitric oxide-mediated vasodilatation by measuring forearm blood flow with venous occlusive plethysmography after administration of acetylcholine (7.5 and 15 micrograms/min) and sodium nitroprusside (3 and 10 micrograms/min), respectively, into the brachial artery. NG-monomethyl-L-arginine was also infused to study the inhibition of basal and stimulated release of nitric oxide. 3. The vasodilatory response to acetylcholine, expressed as area under curve, was significantly decreased in the diabetic patients compared with the control subjects (P = 0.019). NG-monomethyl-L-arginine significantly reduced basal (P < 0.001) and acetylcholine-stimulated blood flow (P < 0.02) in both groups. The vasodilatory response (also expressed as area under curve) to sodium nitroprusside was significantly less (P = 0.044) in the diabetic patients than in the control subjects. 4. In the diabetic patients, impaired vasodilatory responses to acetylcholine were significantly correlated with higher serum triacylglycerols (P = 0.048) and lower high-density lipoprotein-cholesterol concentrations (P = 0.007); the association with high-density lipoprotein was independent of age, glycated haemoglobin and blood pressure. Sodium nitroprusside responses were not correlated with lipid and lipoprotein concentrations. 5. We conclude that there is impaired endothelial and smooth muscle cell function in men with diet-treated non-insulin-dependent diabetes mellitus uncomplicated by overt hypertension or microproteinuria. Endothelial dysfunction may be related to diabetic dyslipidaemia and associated metabolic disturbances.