1. Synchronous interactions between neurons in mesial temporal structures of patients with complex partial seizures were studied using cross-correlation analyses. We recorded spontaneous activity from 293 neurons in 24 patients during the interictal state. Patients had depth microelectrodes chronically implanted in amygdala, hippocampal formation, and parahippocampal gyrus to record epileptic activity. One hundred twenty-five cells were recorded from the temporal lobe commonly initiating seizures (ipsilateral temporal lobe), and 168 cells from the contralateral temporal lobe. Eight hundred forty-three cross-correlograms were constructed between all pairs of simultaneously recorded neurons. Cross-correlogram peaks or troughs that exceeded confidence limits within 200 ms of the origin were considered evidence of synchronous neuronal interaction. 2. Synchronous neuronal interactions were observed in 223 of 843 cross-correlograms. Eighty-six percent of these 223 cross-correlograms showed significant central peaks (peak interactions), suggesting excitatory interactions, whereas the remainder displayed significant central troughs (trough interactions), suggesting inhibitory interactions. 3. Cross-correlograms constructed using cells from the ipsilateral temporal lobe (ipsilateral cross-correlograms) were more likely to display significant central troughs (14/262) than cross-correlograms constructed using cells from the contralateral temporal lobe (6/376; contralateral cross-correlograms). Similarly, cross-correlograms constructed using one cell from each hemisphere (11/205; bilateral cross-correlograms) were also more likely to display significant central troughs (trough interactions) than contralateral cross-correlograms. Both ipsilateral (77/262) and contralateral cross-correlograms (102/376) were more likely to display significant central peaks (peak interactions) than bilateral cross-correlograms (13/205). 4. Cells from different structures in the ipsilateral temporal lobe were more likely to display significant trough interactions (10/ 114) than neurons in different contralateral structures. We also compared the proportion of significant peak interactions between cells within the ipsilateral and contralateral sides of each structure. Neurons in the contralateral entorhinal cortex were more likely to show peak interactions (21/55) than cells from the ipsilateral entorhinal cortex (3/31). Also, cells in the ipsilateral presubiculum showed a higher proportion of peak interactions (9/16) than their contralateral homologues (5/30). 5. Neuronal burst discharges were defined as three or more action potentials (or spikes) separated by interspike intervals of < or = 30 ms, or two spikes separated by an interval of < or = 15 ms. The contribution of burst discharge to synchronous peak interaction was compared between temporal lobes. Cells used to construct ipsilateral cross-correlograms displaying significant central peaks (n = 154) were found to have significantly reduced burst discharge contributions to the observed synchronous peaks in comparison with their contralateral homologues (n = 204). When cross-correlograms were separated by regions, burst discharge contributions to synchronous peak interactions between cells in the ipsilateral hippocampus (n = 72) were significantly smaller than the contributions from cells in the contralateral hippocampus (n = 44). 6. The results suggest that in the interictal state, synchronous neuronal burst discharge is not a distinguishing feature of epileptogenic regions of patients with complex partial seizures, but inhibitory neuronal interactions are increased in regions of seizure initiation. Increases in the strength and spread of local inhibition in seizure initiating regions in these patients may result in a greater proportion of inhibitory interactions and could also cause increased synchrony between isolated action potentials.(ABSTRACT TRUNCATED)