Discrimination of stimulant-responding and nonresponding groups of children with attention-deficit/hyperactivity disorder (ADHD) on the basis of demographic, neurophysiologic, or behavioral variables would be beneficial for clinical and theoretical reasons. Previous researchers have identified many predictor variables, but relationships between predictor and criterion variables generally have been subtle. In addition, few investigations have considered the relative predictive power of the variables. The present study evaluated the multivariate relationship between several predictor variables and response to medication in 336 children with ADHD. Neurologic status, inattention, and overactivity were found to be most likely to predict good response to psychostimulants, whether rated by parents or teachers. Although a number of variables predicted a positive psychostimulant response, the strength of the predictive associations suggests only minimal clinical usefulness.