The clinical and basic literature suggest that hilar cells of the dentate gyrus are damaged after seizures, particularly prolonged and repetitive seizures. Of the cell types within the hilus, it appears that the mossy cell is one of the most vulnerable. Nevertheless, hilar neurons which resemble mossy cells appear in some published reports of animal models of epilepsy, and in some cases of human temporal lobe epilepsy. Therefore, mossy cells may not always be killed after severe, repeated seizures. However, mossy cell survival in these studies was not completely clear because the methods did allow discrimination between mossy cells and other hilar cell types. Furthermore, whether surviving mossy cells might have altered physiology after seizures was not examined. Therefore, intracellular recording and intracellular dye injection were used to characterize hilar cells in hippocampal slices from pilocarpine-treated rats that had status epilepticus and recurrent seizures ('epileptic' rats). For comparison, mossy cells were also recorded from age-matched, saline-injected controls, and pilocarpine-treated rats that failed to develop status epilepticus. Numerous hilar cells with the morphology, axon projection, and membrane properties of mossy cells were recorded in all three experimental groups. Thus, mossy cells can survive severe seizures, and those that survive retain many of their normal characteristics. However, mossy cells from epileptic tissue were distinct from mossy cells of control rats in that they generated spontaneous and evoked epileptiform burst discharges. Area CA3 pyramidal cells also exhibited spontaneous and evoked bursts. Simultaneous intracellular recordings from mossy cells and pyramidal cells demonstrated that their burst discharges were synchronized, with pyramidal cell discharges typically beginning first. From these data we suggest that hilar mossy cells can survive status epilepticus and chronic seizures. The fact that mossy cells have epileptiform bursts, and that they are synchronized with area CA3, suggest a previously unappreciated substrate for hyperexcitability in this animal model.