We studied the effects of age on the limbic and scalp P3 in 45 patients with intracranial electrodes implanted for pre-surgical investigation of focal seizures. Scalp P3 data from a reference group of 24 healthy control subjects were also analyzed for comparison. An auditory oddball paradigm with infrequent stimuli being presented with a probability of 0.20 was used. In normals P3 latency increased by 1.34 msec/year (r = 0.60, P less than 0.01). In the patients limbic and scalp P3 latency increased linearly as a function of increasing age at a rates of 3.85 msec/year (r = 0.58, P less than 0.001) and 2.71 msec/year (r = 0.56, P less than 0.01), respectively. The rate of increase of P3 latency with age was significantly lower in the normal controls, as compared to both the patient scalp (t = 1.79, P less than 0.05) and depth (t = 2.25, P less than 0.005) ERP data. There was no significant difference between the slopes of the patient P3 latency versus age scalp and depth data (t = 1.09, P greater than 0.1). Unlike for normal controls, there was no relationship between age and limbic P3 amplitude (r = 0.02, P greater than 0.1) or age and scalp P3 amplitude (r = 0.17, P greater than 0.1). The differences between controls and patients could be due to: (i) effects of chronic seizures; (ii) long-term effects of anticonvulsant use; (iii) the use of a relatively long inter-stimulus interval which may have selectively affected the patients.(ABSTRACT TRUNCATED AT 250 WORDS)