In the present study, we investigated if the relaxant effects of phosphodiesterase (PDE) III inhibitors on human vessels could be inhibited by a nitric oxide synthase blocker. L-NAME, or by a blocker of ATP-sensitive potassium channels (K-ATP), glibenclamide. The experiments were performed using an isometric myograph in isolated human s.c. small arteries obtained from healthy donors. After a priming procedure consisting of exposure to high potassium (120 mmol litre(-1)) solutions, phenylephrine 10 mu mol litre(-1) and an equilibrium period of 30 min, the preparations were contracted with a thromboxane A(2) mimetic agent, U46619 1 mu mol litre(-1). Subsequently, cumulative concentration-response curves were constructed for the selective PDE III inhibitors amrinone, milrinone and enoximone, and for theophylline and dipyridamole, with and without the addition of L-NAME 100 mu mol litre(-1) or glibenclamide 1 mu mol litre(-1). Addition of L-NAME to the organ bath resulted in significantly higher pEC(50) values (-log of the concentration required for 50% relaxation) for milrinone compared with the control: 2.77 (SEM 0.24) mol litre(-1) (n = 5) vs 3.49 (0.17) mol litre(-1) (n = 7) (P < 0.05). There was no significant difference between any other group. From our data we conclude that the relaxant properties of amrinone, enoximone, theophylline and dipyridamole are not dependent on nitric oxide release or on interaction with K-ATP channels. However, the effect of milrinone may be partly endothelium-dependent in human vessels in vitro.