Arterioles (40–80 micron diameter) were isolated from the hamster cheek pouch, cannulated at both ends, and perfused with 3-(N-morpholino)propanesulfonic acid (MOPS)-buffered physiological salt solution (PSS). The vessels were observed with an inverted microscope and video system, and arteriolar diameter was measured. Arterioles were found to be 100 times more responsive to the alpha 1-adrenoceptor agonist phenylephrine when applied to the adventitial surface than when applied to the luminal surface. In contrast, SKF 89748-A, also an alpha 1-adrenoceptor selective agonist, but with a much greater lipid solubility than phenylephrine, was equipotent from either surface of the arteriole. We hypothesized that the difference between the two drugs was due to the ability of SKF 89748-A to permeate a diffusion barrier in the arteriolar wall because of its lipid-solubility. To test this hypothesis, a spectrum of antagonists with different sites of action and lipid solubilities was tested. The alpha-adrenoceptor antagonists phentolamine and benextramine and the muscarinic receptor antagonists atropine, scopolamine, and methscopolamine were all found to be more potent at blocking the action of appropriate agonists when applied to the same surface of the arteriole as the agonist than when applied to the opposite surface. Octanol-water partition coefficients were measured for each of the compounds, and these were found to be highly correlated with the ratio of luminal potency to adventitial potency for each of the drugs tested. These data support the hypothesis that the endothelial cell layer in these arterioles forms a barrier to the diffusion of small, water-soluble molecules from the lumen to the smooth muscle cell layer. Such a barrier may have a significant effect on arteriolar reactivity.