Nitric oxide is an endogenous vasodilator produced from L-arginine and oxygen by stereospecific enzymes. L-arginine itself can act as a vasodilator when administered at high doses to humans. This effect has been attributed by some to provision of extra substrate for production of nitric oxide. This work compares L-arginine-induced vasodilation with that caused by D-arginine, hyperosmolar sodium chloride and by other cationic amino acids in the resting forearm vasculature of normal subjects. By these means we identify whether L-arginine-induced vasodilation has a component related to stereospecific provision of substrate for nitric oxide production, or whether it can be accounted for by other phenomena. Effects of hyperosmolar sodium chloride and both L and D isomers of arginine, lysine and ornithine on forearm blood flow in eight normal male subjects were compared by bilateral forearm venous occlusion plethysmography. Vasodilator responses to saline and each amino acid were compared as the area under dose-response curves with single-factor analysis of variance (ANOVA). The magnitude of vasodilation obtained with the D isomers of each amino acid was compared with the L counterpart by application of single-factor ANOVA to the appropriate areas under dose-response curves. All three cationic amino acids increased forearm blood flow. Part of the increase could be related to the high osmolality of infusates-comparison with equiosmolar sodium chloride solutions shows that ornithine does not differ significantly as a vasodilator (P > 0.4), but that arginine and lysine have greater vasodilator effects than can be accounted for by osmolality alone (P < 0.001). The D isomers of arginine and lysine were more potent dilators than their L counterparts (arginine P < 0.03, lysine P < 0.02). The vasodilator effects of arginine in the forearm vascular bed at rest are not stereospecific, they are common to other cationic amino acids, greater for D isomers and occur only when normal plasma concentrations are raised far above the physiological range. These features suggest that the vasodilator effect of arginine is in part physical and related to the presence of the molecule in the vessel lumen. They do not suggest that increased provision of substrate, with a consequence of increased nitric oxide production, is the principal basis of L-arginine-induced vasodilation in normal humans.