Despite optimal treatment, 30% of epilepsy patients develop intractable epilepsy and continue to have recurrent seizures or other symptoms of epileptic syndrome restricting their ability to lead a full life. Hippocampal sclerosis is found in 60-70% of patients with intractable temporal lobe epilepsy (TLE). However, it is not known whether the damage in the hippocampus is the cause or the consequence of TLE. The purpose of the present series of studies was to investigate with magnetic resonance imaging (MRI) the appearance of medial temporal lobe damage during the course of partial epilepsy, and, particularly, to determine whether recurrent or prolonged seizures contribute to the damage. Altogether 259 partial epilepsy patients were investigated with quantitative MRI. High lifetime seizure number, complex febrile convulsions in the medical history, and early age at the onset of spontaneous seizures contributed to hippocampal damage in patients with TLE. The risk factors that predicted amygdaloid volume reduction were intracranial infection and complex febrile convulsions. Damage in the hippocampus or in the amygdala was rare at the time of first spontaneous seizures in TLE. In contrast, hippocampal damage was apparent in chronic TLE patients with years of frequent seizures. Chronic cryptogenic drug-resistant TLE patients had smaller mean hippocampal volumes ipsilateral to the seizure focus than controls. In all TLE patients, ipsilateral hippocampal volume correlated negatively with the lifetime seizure number. The mean amygdaloid volumes in chronic TLE patients did not differ from those in controls. However, about 20% of chronic patients had > or = 20% volume reduction in the amygdala. The mean volumes of the entorhinal cortex ipsilateral to the epileptic focus in cryptogenic TLE patients did not differ from those in controls. However, the entorhinal cortex was damaged in a subpopulation of TLE patients with associated hippocampal damage TLE. The findings of the present series of studies support the hypothesis that damage in the medial temporal lobe structures may be both the cause and consequence of TLE. The data provide evidence that in some patients hippocampal damage may progress as a function of repeated seizures, and argue for efficient drug therapy or early surgery to reach complete seizure control. Future research should address strategies for disease-modifying therapies and ultimately remission of the epileptic process.