Despite its potent proconvulsant effects in vitro, the adenosine A1 receptor antagonist 1,3-dipropyl-8-cyclopentylxanthine (DPCPX) does not induce seizures when administered in vivo. This contrasts with the effects of less selective adenosine antagonists such as theophylline or cyclopentlytheophylline, and led us to reexamine the nature of DPCPX-induced epileptiform activity. In the present study, we report that proconvulsant effects of bath-applied DPCPX in rat hippocampal slices are only observed after a preceding stimulus such as NMDA receptor activation or brief tetanic stimulation. While this may be due to the absence of a basal "purinergic tone", the relatively high interstitial concentrations of adenosine present in the slice suggest that access of the drug to A1 receptors may instead be prevented by tightly coupled endogenous adenosine, with the ternary adenosine-A1 receptor-G protein complex stabilised in the high-affinity conformation by a coupling cofactor. This implies that a substantial percentage of adenosine A1 receptors are inactive under physiological conditions, but that access of adenosine A1 receptor antagonists may be facilitated under pathological conditions. Once induced, DPCPX-evoked spiking persists for long periods of time. A "kindling" effect of A1 receptor blockade is unlikely, since persistent spiking is not usually observed with less selective A1 antagonists even after prolonged application. Alternatively, endogenous adenosine released during increased neuronal activity may activate A2 receptors during selective A1 blockade. The most important factor determining the duration of DPCPX-induced spiking, however, may be a persistence of the drug in the tissue and subsequent access to the A1 receptor via a membrane-delineated pathway, since DPCPX-induced spiking could be shown to decrease markedly after a transient superfusion of theophylline. This hypothesis, which implies that the apparent affinity of adenosine antagonists for the A1 receptor is in part a function of their membrane partitioning coefficient, is supported by a close correlation between alkylxanthine logP values obtained from the literature and their Ki value at A1 receptors, but not at the enzyme phosphodiesterase, whose xanthine binding site is presented to the cytosol. The implications for the therapeutic value of purinergic drugs are discussed.