Identifying the role of GABA neurons in the development of an epileptic state has been particularly difficult in acquired epilepsy, in part because of the multiple changes that occur in such conditions. Although once questioned, there is now considerable evidence for loss of GABA neurons in multiple brain regions in models of acquired epilepsy. This loss can affect several cell types, including both somatostatin- and parvalbumin-expressing interneurons, and the cell type that is most severely affected can vary among brain regions and models. Because of the diversity of GABA neurons in the hippocampus and cerebral cortex, resulting functional deficits are unlikely to be compensated fully by remaining GABA neurons of other subtypes. The fundamental importance of GABA neuron loss in epilepsy is supported by findings in genetic mouse models in which GABA neurons appear to be decreased relatively selectively, and increased seizure susceptibility and spontaneous seizures develop. Alterations in remaining GABA neurons also occur in acquired epilepsy. These include alterations in inputs or receptors that could impair function, as well as morphological reorganization of GABAergic axons and their synaptic connections. Such axonal sprouting could be compensatory if normal circuits are reestablished, but the creation of aberrant circuitry could contribute to an epileptic condition. The functional effects of GABA neuron alterations thus may include not only reductions in GABAergic inhibition but also excessive neuronal synchrony and, potentially, depolarizing GABAergic influences. The combination of GABA neuron loss and alterations in remaining GABA neurons provides likely, though still unproven, substrates for the epileptic state.