Thirteen boys with Attention-Deficit Hyperactivity Disorder (ADHD) completed 80 arithmetic problems presented on a computer screen by typing a two digit answer. On half the trials, a foot press was required to terminate a computer-generated tone presented 2 seconds before, 1 second before, 1 second after, and 2 seconds after arithmetic problem presentation. Compared to placebo, methylphenidate resulted in significantly faster reaction times (RTs) to tone probes and faster answers to arithmetic problems when the two tasks did not overlap in time, but not when simultaneous processing was required when the probe was presented 2 seconds after arithmetic problems. When dual processing taxed cognitive capacity, methylphenidate still improved accuracy on the primary arithmetic task relative to placebo, but at the expense of speed of performance on the secondary RT task. When ADHD children fail to allocate available resources to a primary cognitive task, treatment with methylphenidate may result in reallocation of existing cognitive capacity from a secondary task to the primary task.