OBJECTIVE To test the hypothesis that insulin has a direct vasodilatory effect on retinal arteries and their branches and to investigate the mechanisms involved. METHODS Segments of porcine retinal arteries were dissected, cannulated, and perfused. Vessel diameter was measured continuously on-line. Vessels were precontracted to 66% +/- 0.9% (SEM, n = 148) of their original diameter by perfusing with 124 mM K(+)-Krebs solution. Dose-response curves to insulin (2 to 2000 microU/ml) were compared for extraluminal (EL), intraluminal (IL), and combined IL-EL application. The effect of cyclooxygenase and nitric oxide synthase inhibition on the insulin response was determined, as was Ca2+ channel involvement. RESULTS EL insulin alone had no significant effect on vessel diameter. IL insulin produced a dose-dependent dilatation of 5.6% +/- 2.9% (n = 22) of the K+ contracted diameter at 200 microU/ml and up to 12.4% +/- 3.6% (n = 22) by 2000 microU/ml, whereas combined IL-EL insulin application caused dilatation at all concentrations, rising to 15.1% +/- 2.9% (n = 44) at 200 microU/ml and 19.7% +/- 3% (n = 44) at 2000 microU/ml. IL indomethacin (5 x 10(-5) M) had no significant effect on the insulin-induced dilatation, whereas IL L-NAME (10(-4) M) inhibited insulin dilatation completely. The addition of EL verapamil (10(-6) M) during insulin-induced dilatation resulted in further dilatation to 37.8% +/- 4.2% (n = 18). However, the addition of insulin to verapamil-dilated vessels caused no further dilatation. Exposure to EL insulin while the IL K+ contraction dose-response curve was measured had no effect. Results in main arteries and branches did not differ. CONCLUSIONS The IL application of insulin dilates potassium-contracted pig retinal arteries. This effect was enhanced by the EL presence of insulin, which did not result in dilatation when it was administered alone. The dilatation response was mediated by nitric oxide but not by prostaglandins. There was some evidence for the involvement of Ca2+ channels in insulin-induced dilatation. These results imply that insulin is a vascular regulator in normal conditions and may have relevance to the vascular changes occurring in diabetes and hypertension in the retina.