We have utilized the favorable signal-to-noise ratios provided by whole- cell recording, combined with variance analysis, to determine the pre- or postsynaptic actions of a variety of manipulations on unitary EPSPs evoked by low-intensity stimulation of afferents to CA1 pyramidal neurons in slices of hippocampus. Estimates of quantal content (mcv) were determined by calculating the ratio of the squared average unitary EPSP amplitude (determined from 150–275 responses) to the variance of these responses (M2/sigma 2), while quantal amplitudes (qcv) were estimated by calculating the ratio of the response variance to average EPSP size (sigma 2/M). Estimates of mcv were highly correlated with those determined using the method of failures (mf). With paired stimulation (50 msec interpulse interval) there was a significant facilitation of the second unitary EPSP, accompanied by an increase in mcv, but not qcv, suggesting that this facilitation was of presynaptic origin. Superfusion of hippocampal slices with various concentrations of adenosine, the A1-selective adenosine receptor agonist cyclohexyladenosine, or the Ca2+ channel blocker cadmium significantly reduced average unitary EPSP amplitudes and mcv, without significantly altering qcv, suggesting a presynaptic locus for this inhibition. The 50% effective concentration for the apparent presynaptic action of adenosine on mcv in the present study (5.7 microM; 95% confidence limits = 4.2–7.7 microM) was significantly lower than its EC50 for reducing conventional, large EPSPs (33 microM; recorded with high- resistance microelectrodes), or extracellular field EPSPs (29 microM), as previously reported by this laboratory. The glutamate receptor antagonist 6,7-dinitroquinoxaline-2,3-dione (DNQX) reduced average unitary EPSP amplitudes; in contrast to the above manipulations, it had no effect on mcv, but significantly altered qcv, which is consistent with its presumed postsynaptic mechanism of action. We conclude from these data that adenosine presynaptically reduces synaptic strength at Schaffer collateral-commissural synapses in the hippocampus by diminishing the number of quanta released, not by reducing the size of these individual quanta or postsynaptic sensitivity to excitatory neurotransmitter. These results suggest that the mechanism by which adenosine inhibits synaptic transmission in the hippocampus is similar, if not identical, to the mechanism by which it inhibits synaptic transmission at the neuromuscular junction.