The adenosine concentration of exudate formed on the surface of isolated perfused rat hearts has been used to obtain estimates of interstitial values. At a constant perfusion of approximately 15 ml/min/g, exudate was collected from below ring seals that either fitted snugly (compressing seals) or that acted as wicks (wicking seals) to deflect venous effluent away from the apical surface. Steady state exudates flows obtained below each of these seals were 0.96 +/- 0.05 ml/min and 0.18 +/- 0.02 ml/min, respectively. Adenosine concentration of surface exudate and venous effluent from hearts with the compressing seal were 130 +/- 8 nM and 23 +/- 3 nM, respectively, and from those with the wicking seal were 770 +/- 93 nM and 36 +/- 9 nM, respectively. Interstitial adenosine concentration in a situation with no net filtration may be slightly higher than that achieved in the exudate from preparations with the wicking seal. Addition of exogenous adenosine to the perfusate (1.0 microM) decreased vascular resistance and automaticity of all preparations, increased the venous effluent adenosine concentration to 236 +/- 18 nM and 251 +/- 30 nM with the compressing and wicking seals, respectively, but did not significantly alter the exudate adenosine concentration with either of the seals. This finding suggests that increases in vascular adenosine may influence functional characteristics without altering interstitial levels. Perfusion with 10 microM adenosine increased adenosine concentration in both effluent and exudate in all preparations but the gradient was reversed so that effluent levels were significantly higher than exudate levels. We conclude that venous adenosine determinations significantly underestimate the interstitial adenosine concentration associated with endogenous adenosine production and significantly overestimate the interstitial levels achieved by infusion of exogenous adenosine.