In vitro experiments have shown that luminally applied water-soluble vasoactive materials have limited access to arteriolar smooth muscle cells, and as a result, the responses to such agents applied luminally are less than the responses to those applied adventitially. To determine the extent to which this ‘compartmentation’ influences arteriolar responsiveness to blood-borne water-soluble vasoconstrictors in vivo, we applied phenylephrine, vasopressin and angiotensin II to arterioles in the hamster cheek pouch both by luminal perfusion, and by topical application to the arteriolar smooth muscle via micropipettes. The arterioles were about 2 orders of magnitude more sensitive to these water-soluble vasoconstrictors when they were applied topically than when they were applied luminally. In contrast, the arterioles were almost equally sensitive to the lipid-soluble α<sub>1</sub>-adrenoceptor agonist SKF 89748-A applied by either route. The venular wall appears to be much less effective as a barrier than the arteriolar endothelium. Phenylephrine and vasopressin both elicited large arteriolar constrictions when perfused through venules in close proximity to the arteriole, and these constrictions were larger than those observed when the drug was applied to the arteriole’s own lumen. Our observations confirm that the arteriolar endothelium can inhibit the direct access of water-soluble blood-borne agents to the arteriolar smooth muscle in vivo, and they suggest that the capillaries and venules could be the primary routes of access for water-soluble agents from the blood to the arteriolar smooth muscle.