1. Voltage-clamp recording of Ca(2+)-activated chloride currents in Xenopus oocytes was used to study the effects of caffeine on the liberation of intracellular Ca2+ induced by photo-release of inositol 1,4,5-trisphosphate (InsP3) from caged InsP3. Bath application of caffeine, at concentrations between 0.1 and 10 mM, reduced or abolished the current evoked by photo-release of InsP3 and by microinjection of InsP3. 2. Caffeine did not appreciably reduce currents evoked by injection of Ca2+ into oocytes, whereas measurements using the Ca2+ indicator Rhod-2 showed that it instead inhibited the liberation of Ca2+ by InsP3. 3. Caffeine increased the threshold amount of InsP3 required to evoke a current response and proportionally reduced the currents evoked by suprathreshold levels of InsP3. 4. Theophylline and 3-isobutyl-1-methylxanthine (IBMX) were much less potent than caffeine, and few changes were seen in the InsP3 responses following application of forskolin or intracellular injection of cyclic AMP. Thus, inhibition of InsP3 responses by caffeine does not arise through inhibition of phosphodiesterase enzymes. 5. Even at high (10 mM) concentrations, caffeine did not itself elicit any clear Ca(2+)-activated current. It is therefore unlikely that inhibition of the InsP3 responses arise because caffeine itself liberates Ca2+ from intracellular stores. 6. The site of action of caffeine is intracellular, because injections of caffeine into the oocyte strongly inhibited responses to InsP3, whereas local extracellular applications of similar amounts were almost without effect.