To study the vasodilatory capacity of subendocardial (ENDO) arterioles, we evaluated the reactive hyperemic responses of ENDO as well as subepicardial (EPI) arterioles in 40 dogs by our needle-probe intravital microscope. We also examined the individual and combined effects of an ATP-sensitive K+ channel blocker (glibenclamide, 200 micrograms/kg), an inhibitor of nitric oxide synthase (NG-monomethyl-L-arginine [L-NMMA], 2 mumol/min, 20 minutes), and an adenosine-receptor antagonist (8-phenyltheophylline [8PT], 0.75 mumol/min, 15 minutes). The percent increase in end-diastolic diameter of ENDO arterioles was larger (P < .01) than that of EPI arterioles during reactive hyperemia, especially for the arterioles larger than 120 microns (P < .01). The diastolic-to-systolic vascular pulsation amplitude at the peak flow was greater in ENDO than EPI arterioles (25% versus 6%, P < .05). Compared with control conditions, the presence of both glibenclamide and L-NMMA suppressed the vasodilation responses of ENDO arterioles (P < .01 for both) and EPI arterioles (P < .05 for both). The effect of L-NMMA was greater in ENDO arterioles (P < .01), but that of glibenclamide was not different between ENDO and EPI arterioles. 8PT influenced the hyperemic response, although statistical significance was found only in the flow response. The effect of combined infusion of L-NMMA and glibenclamide with or without 8PT was greater than that of individual infusions in both ENDO and EPI arterioles. Conclusions are as follows: (1) The vasodilatory response of ENDO arterioles was even larger than that of EPI arterioles. Thus, the smaller flow reserve of ENDO arterioles may be caused by other factors, including the greater effects of myocardial compression and nitric oxide on the ENDO arterioles. (2) The vascular responses of ENDO and EPI arterioles were modulated by both endothelium-independent and -dependent vasodilative factors, and the effect of each factor including adenosine was associated with the effects of others.