A study was performed to examine the utility of an ERP-based irrelevant probe technique for the assessment of variations in mental workload. Ten highly trained Navy radar operators performed a simulated radar-monitoring task which varied in the density and type of targets to be detected and identified. This task was performed in the presence of a series of irrelevant auditory probes which the radar operators were instructed to ignore. Prior to performing the radar-monitoring task the subjects performed a block of auditory detection trials in which they were asked to respond to the occurrence of one of two low probability tones and ignore the other low probability tone along with a higher probability standard tone. ERPs were recorded from the occurrence of the tones in both the baseline and low and high workload radar-monitoring conditions. The amplitude of the N100, N200, and early and late mismatch negativity (MMN) components decreased from the baseline to the low load radar-monitoring task and again with an increase in the difficulty of the radar-monitoring task. P300 amplitude was sensitive only to the introduction of the radar-monitoring task. These results are interpreted with respect to the phenomenon of attentional capture and suggest that the ERP-based irrelevant-probe technique might prove an effective method for the nonintrusive evaluation of increases in mental workload in complex tasks.