We studied scalp-recorded auditory event-related potentials (ERPs) of 30 untreated patients with new-onset temporal lobe epilepsy and 30 age- and sex-matched normal controls. This study was designed to eliminate the effects of intractability of seizures and chronic use of antiepileptic drugs on P300 auditory ERPs. There were no statistically significant differences in both latency and amplitude of P300 between the two groups. Similar methods were also used to analyze component latencies and amplitudes of ERPs of 9 patients who had hippocampal sclerosis with comparison to control subjects. There were no statistically significant differences between these two groups as well. Our study evidently does not support temporal lobe sources of P300 scalp-recorded auditory ERPs. We also conclude that the scalp-recorded auditory ERPs procedure is not a useful tool to evaluate temporal lobe epilepsy.