To design useful experimental models of epilepsy, it is necessary to clearly understand the known clinical-pathologic features of the disease process. Studies of mesial temporal lobe epilepsy (MTLE) patients have identified several distinctive clinical and pathophysiologic characteristics and many of these can be analyzed in experimental models. For example, patients with typical MTLE have medical histories that often contain an initial precipitating injury (IPI), are likely to have hippocampal sclerosis in the surgical specimen, and have better seizure outcomes than patients with typical idiopathic temporal seizures (i.e. cryptogenic). Hippocampal from children as young as age 1 year with IPI histories also demonstrate neuron damage similar to adults with hippocampal sclerosis. Compared to IPI patients without seizures (i.e. trauma, hypoxia, etc.), IPI cases with severe seizures showed younger ages at the IPI, shorter latent periods, and longer durations of habitual MTLE. Hippocampal damage is often bilateral, however, the epileptogenic side shows hippocampal sclerosis and the opposite side usually shows only mild neuron losses. Moreover, MTLE patients show declines in hippocampal neuron densities with very long histories of habitual seizures (15 to 20 years), however, the additional neuron loss adds to the template of hippocampal sclerosis and occurs in limited subfields (granule cells, CA1 and prosubiculum). Hippocampal axon and synaptic reorganization is another pathologic feature of MTLE, and involves granule cell mossy fibers and axons immunoreactive for neuropeptide upsilon, somatostatin, and glutamate decarboxylase (which synthesizes GABA). Finally, MTLE patients with hippocampal sclerosis show increased granule cell mRNA levels for brain derived neurotropic factor, nerve growth factor, and neurotrophin-3 that correlate with mossy fiber sprouting or with declines in Ammon's horn neuron densities. Taken together, our data support the following concepts: (1) The pathogenesis of MTLE is associated with IPI histories that probably injure the hippocampus at some time prior to habitual seizure onsets, (2) most of the damage seems to occur with the IPI, (3) there can be additional neuron loss associated with long histories, (4) another pathologic feature of MTLE is axon reorganization of surviving fascia dentata and hippocampal neurons, and (5) reorganized axon circuits probably contribute to seizure or propagation.