The effect of mild head injury on event-related potential (ERP) correlates of Stroop task performance was explored with the aim of further elucidating the basis of processing impairments after mild head injury. Computer- and card-based Stroop tasks were employed to assess attention function. A sequence of incongruent color words were presented followed by a sequence of congruent color words (printed in congruent colors). Control performance was equivalent on computer- and card-based versions of the incongruent task and faster on the congruent card task than the preceding congruent computer task. The mild head injury group were as fast as controls on the computer-based task but made more errors. However, they were relatively slower on both the congruent and incongruent parts of the card-based task and made more errors in the incongruent task. ERP correlates of computer-based Stroop task performance suggested a greater allocation of attention resources in the incongruent condition in both groups in the form of relatively greater negativity in the latency range 350 to 450 ms with a distribution consistent with the activation of the anterior cingulate gyrus. In addition the mild head injured group showed relatively greater enhancement than the control group in this latency range in both congruent and incongruent conditions. There was, however no evidence of reduced amplitude P1, N1, N2, or P3b deflections. Trails, digit symbol, digit span and auditory verbal learning tests (AVLT) were also administered. Mild head injured participants were impaired only on the AVLT. The finding of greater ERP negativity in the mild head injured group is consistent with greater allocation of attention resources to achieve equivalent performance in the computer-based Stroop task. (JINS, 2002, 8, 828–837.)