Non-modulation has been proposed as an intermediate phenotype in human essential hypertension. The trait is characterized by blunted aldosterone and renal plasma flow responses to short-term angiotensin II (Ang II) infusion. Elevated tissue Ang II levels or decreased tissue adenosine levels could account for this decreased sensitivity to Ang II. In support of the latter possibility, endogenous adenosine has been shown to contribute to the renal vasoconstrictive response to Ang II in animals. We therefore tested the hypothesis that endogenous adenosine contributes to modulation of renal plasma flow in sodium-replete humans. We examined the effect of long-term administration of the adenosine receptor antagonist caffeine on baseline renal plasma flow and on the renal plasma flow response to short-term Ang II infusion in six salt-replete normotensive subjects in a single-blind, placebo-controlled study. para-Aminohlppurate clearance was used to assess renal plasma flow. Ang II was infused in graded doses (03 to 3 ng/kg per minute) in the presence and absence of caffeine (250 mg PO TID for 7 days). Blood pressure, plasma renin activity, Ang II, electrolytes, and pora-aminohippurate clearance were measured before and after each dose of Ang II. Caffeine did not alter either baseline blood pressure or the blood pressure response to Ang II but did increase baseline plasma renin activity from 0.72±0.09 to 1.42±0.26 ng angiotensin I/mL per hour (P=.01). Caffeine decreased the baseline renal plasma flow from 553±38 to 476±31 mL/min per 1.73 m 1 (P=.OO4) and attenuated the renal plasma flow response to a 3 ng/kg per minute infusion of Ang II (-106.5±25.2 versus -170.5±18 mL/min per 1.73 m 2 , P=.0O6). These data demonstrate that caffeine modulates baseline renal plasma flow and the renal plasma flow response to exogenous Ang II and therefore support the hypothesis that adenosine contributes to modulation of renal plasma flow in salt-replete humans. Thus, non-modulation may be partially acquired, and caffeine consumption must be controlled in studies that define modulation phenotype. (Hypertension. 1993;22:847-852.)