This study determined the levels of adenosine in the renal medullary interstitium using microdialysis and fluorescence HPLC techniques and examined the role of endogenous adenosine in the control of medullary blood flow and sodium excretion by infusing the specific adenosine receptor antagonists or agonists into the renal medulla of anesthetized Sprague-Dawley rats. Renal cortical and medullary blood flows were measured using laser-Doppler flowmetry. Analysis of microdialyzed samples showed that the adenosine concentration in the renal medullary interstitial dialysate averaged 212 +/- 5.2 nM, which was significantly higher than 55.6 +/- 5.3 nM in the renal cortex (n = 9). Renal medullary interstitial infusion of a selective A1 antagonist, 8-cyclopentyl-1,3-dipropylxanthine (DPCPX; 300 pmol. kg-1. min-1, n = 8), did not alter renal blood flows, but increased urine flow by 37% and sodium excretion by 42%. In contrast, renal medullary infusion of the selective A2 receptor blocker 3, 7-dimethyl-1-propargylxanthine (DMPX; 150 pmol. kg-1. min-1, n = 9) decreased outer medullary blood flow (OMBF) by 28%, inner medullary blood flows (IMBF) by 21%, and sodium excretion by 35%. Renal medullary interstitial infusion of adenosine produced a dose-dependent increase in OMBF, IMBF, urine flow, and sodium excretion at doses from 3 to 300 pmol. kg-1. min-1 (n = 7). These effects of adenosine were markedly attenuated by the pretreatment of DMPX, but unaltered by DPCPX. Infusion of a selective A3 receptor agonist, N6-benzyl-5'-(N-ethylcarbonxamido)adenosine (300 pmol. kg-1. min-1, n = 6) into the renal medulla had no effect on medullary blood flows or renal function. Glomerular filtration rate and arterial pressure were not changed by medullary infusion of any drugs. Our results indicate that endogenous medullary adenosine at physiological concentrations serves to dilate medullary vessels via A2 receptors, resulting in a natriuretic response that overrides the tubular A1 receptor-mediated antinatriuretic effects.