Thirty-two children with attention deficit hyperactivity disorder (ADHD) undergoing a 4 week double-blind medication assessment (methylphenidate) and 32 normal controls were studied using event-related potentials (ERPs). The ERPs were recorded from 13 active electrodes during a visual feature detection task. Significant age effects were found in N2, P3a and P3b latencies, that did not interact with group. The P3a and P3b latencies were significantly longer in the ADHD children on baseline testing; there were no latency differences between the groups of children when the normal controls were compared with the ADHD children on their optimal drug dosage (as determined by extensive behavioural and cognitive assessments). There were no significant distributional effects either between groups, or with the ADHD children as a function of medication; there were also no significant differences in reaction time measures. Thus, only the ERPs reflected slowed processing in the ADHD children that normalized on appropriate medication.