In the present study, three different types of probe evoked potential (EP) techniques were examined using a mental math task with three levels of difficulty. One probe condition consisted of presenting flashes at 5 sec intervals during the performance of each task level. The other conditions were designed such that probe flashes were presented at 250 and 750 msec after the onset of each mental math task item. Baseline (no task) measurements were taken for all three probe conditions. Subjects were 6 males and 4 females who participated in an earlier study (Yolton, Wilson, Davis and McCloskey, 1987), and were recalled for the present experiment. Results of the RT data replicated those found in Yolton, et al (1987), where RT increased as task level increased. The EPs obtained from each of the probe conditions showed different patterns of variation with task demand. The 5 sec probe showed differences between the no-task baseline and all other levels of the task, but not between task levels. The 250 msec probe EPs showed graded changes with task level, whereas the 750 msec probe EPs were similar to those found with the 5 sec probe EPs. These measures support the notion that probe EPs are not only measures of tonic activity, but are also measures of phasic activity. To index phasic activity, probes must be present during relevant times during task performance. The usefulness of this technique is discussed in terms of moment-to-moment fluctuations of processing demands in applied situations.