To analyse the effect of epilepsy an P300 event-related potentials we studied 27 patients with idiopathic generalized epilepsy (IGE), 13 patients with temporal lobe epilepsy (TLE) and 60 normal controls. The prolongation of P300 latencies was highly cor related with increasing age in controls but not epileptic patients. The age-corrected P300 latency used in this study was actual P300 latency-predicted P300 latency (predicted P300 (msec)=306.20+0.79 age, P=0.001, R2=0.32). By using ANOVA analysis, the age-corrected latencies of P300 were significantly longer in TLE patients (19.72+/-47.82 msec, mean+/-SD) than in IGE patients (10.97+/-36.97 msec) and controls (0.23+/-20.28 msec). Likewise, significantly prolonged P300 latencies were seen in the epileptic patients with a seizure frequency more than 400 times (37.21+/-47.50 msec). The multivariate-adjusted odds ratio for those who had TLE was 10.97 (95% CI=3.99 - 30.14 ) in the prolonged latencies of P300 compared with that of IGE patients. The odds ratio of longer latencies of P300 was 7.43 (95% CI=2.75 - 20.08) among those who had a high seizure frequency (> or =400 times) compared with those who had a low seizure frequency. No interaction between TLE and high frequency of attacks was found. The age at onset of seizure and duration of illness was not associated with P300 latency prolongation. From the above results, we might infer that the seizure type of TLE and a high frequency of seizure are two major independent precipitate factors for abnormal latencies of P300 in the epileptic patients.