We tested the hypothesis that metformin produces arterial dilatation indirectly, by directly exposing the endothelial surface, of an occluded test segment of the pig iliac artery in vivo, to test blood containing metformin or excess insulin, with and without the presence of the nitric oxide (NO) synthase inhibitor NG-nitro-L-arginine methyl ester hydrochloride. Such exposure to metformin 1 mu g/mL caused the artery to dilate at constant pressure, and this was abolished when NG-nitro-L-arginine methyl ester hydrochloride was coadministered with metformin. The onset of dilatation occurred approximately 4 minutes after the commencement of endothelial exposure to metformin; this contrasts with the approximate 10 minutes required for a similar response to luminal hyperinsulinemia. After the release of flow occlusion, the subsequent flow-mediated dilatation was slightly but significantly enhanced compared with control for metformin; the effect of insulin on flow-mediated dilatation was not statistically significant. The hypothesis was disproved, as we have shown that insulin and metformin, like insulin, directly stimulate NO production by endothelium of a conduit artery; this function may be of value in delaying the atherothrombotic process. The time taken for the commencement of NO production is shorter for metformin than for insulin; the clinical relevance of this finding is unclear.