Different instructional sets for moderate vs. high response speeds were given in a visual letter-matching task. On 50 percent of the trials an auditory probe requiring a simple RT response was presented in one of four temporal locations within the ongoing visual task. Probe locations were selected to maximize differences in the processing demands of the dual, auditory-visual task (see Posner and Boies, 1971). The N100 of the event-related potential associated with the auditory probe was larger under the high response speed set than under the moderate set, regardless of probe location. P300 amplitude was also greater under the high speed set relative to the moderate set but only when the task demands were high. The results were interpreted in terms of the interaction between arousal and task demands and a neural inhibition model of the P300 was discussed.