P300 is a cognitive evoked potential that evaluates attention and information processing. This study uses auditory and visual P300 topography to develop a classification of attention deficit disorder (ADD), and to find predictors of treatment response to the stimulant pemoline. Forty-five ADD children ages 6 to 15 were administered auditory and visual P300 using 31 scalp electrodes. They were compared with 39 normals. Patients were treated with pemoline, and good and poor responders compared. There were no P300 differences between normals and ADD patients. Good and poor responders to pemoline were clinically identical. Poor pemoline responders had smaller right fronto-central auditory P300 amplitudes than good responders. The ratio of right fronto-central to parietal auditory P300 amplitude, had a sensitivity of 0.70 and specificity of 0.76, as a test for good pemoline response. A ratio greater than 0.5 predicted good response to pemoline, while a ratio less than 0.5 predicted poor response. Treatment with pemoline produced no P300 changes. We conclude that P300 topography classifies ADD into group 1 with normal P300 topography and good response to pemoline, and group 2 with small right fronto-central auditory P300 amplitudes and poor response to pemoline.