The "dormant basket cell" hypothesis suggests that postinjury hippocampal network hyperexcitability results from the loss of vulnerable neurons that normally excite insult-resistant inhibitory basket cells. We have reexamined the experimental basis of this hypothesis in light of reports that excitatory hilar mossy cells are not consistently vulnerable and inhibitory basket cells are not consistently seizure resistant. Prolonged afferent stimulation that reliably evoked granule cell discharges always produced extensive hilar neuron degeneration and immediate granule cell disinhibition. Conversely, kainic acid-induced status epilepticus in chronically implanted animals produced similarly extensive hilar cell loss and immediate granule cell disinhibition, but only when granule cells discharged continuously during status epilepticus. In both preparations, electron microscopy revealed degeneration of presynaptic terminals forming asymmetrical synapses in the mossy cell target zone, including some terminating on gamma-aminobutyric acid-immunoreactive elements, but no evidence of axosomatic or axoaxonic degeneration in the adjacent granule cell layer. Although parvalbumin immunocytochemistry and in situ hybridization revealed decreased staining, this apparently was due to altered parvalbumin expression rather than basket cell death, because substance P receptor-positive interneurons, some of which contained residual parvalbumin immunoreactivity, survived. These results confirm the inherent vulnerability of dendritically projecting hilar mossy cells and interneurons and the relative resistance of dentate inhibitory basket and chandelier cells that target granule cell somata. The variability of hippocampal cell loss after status epilepticus suggests that altered hippocampal structure and function cannot be assumed to cause the spontaneous seizures that develop in these animals and highlights the importance of confirming hippocampal pathology and pathophysiology in vivo in each case.