Quantal analysis has been applied to the inhibitory synapses made by single spiking local interneurons onto several nonspiking local interneurons (and motorneurons) in the locust CNS. Transmission at these synapses appears to be mediated by GABA. The apparent reversal potential of the IPSP and inhibitory postsynaptic current were -80 to - 85 mV, a value similar to that of the potential evoked by pressure- applied GABA. This reversal potential was 25–30 mV more negative than the resting potential of the nonspiking interneurons in the experimental conditions. The statistical properties of release at these synapses were studied by recording simultaneously from pre- and postsynaptic interneurons with intracellular electrodes. The distribution of postsynaptic potential amplitudes could be described by a simple binomial model, implying uniformity of binomial p (probability of release at a single release site) for each synapse. The mean quantal amplitude was 290 +/- 110 microV, and the mean quantal content m of the IPSPs was 6.25 +/- 2.83. The mean values of binomial n (average size of the releasable pool) and p were 13.11 +/- 2.8 and 0.45 +/- 0.16, respectively. Numerical simulations of statistical experiments were performed to test whether the IPSP amplitude distribution histograms might be misleadingly indicative of quantal release. These simulations showed that such a hypothesis was very unlikely. When a spiking local interneuron was impaled, several of its target interneurons could sometimes be successively sampled. Quantal analysis was then performed with the different IPSPs evoked, in identical conditions, by a same presynaptic interneuron, and the quantal parameters were compared between the synapses. It was found that binomials n and p and their product m generally differed between the synapses made by a given spiking interneuron onto different target neurons. These results show that quantal contents can vary for the many synapses made centrally by one interneuron, and suggest that this variability may arise from differences in release probabilities between the sites associated with different synapses.