This experiment examined the effect of physical exercise on measures of cognitive performance, Raven's Matrices, and an adaptation of the revised WAIS arithmetic subtest. We also tested the inverted-U hypothesis of an interactive relation between exercise-induced arousal and cognitive performance. 50 physically active men were assigned to five groups (n = 10) of equal physical fitness based on predicted maximum oxygen uptake. Three exercise groups undertook bench stepping at mean power outputs of either 47, 75, or 120 watts. One control group played Bingo and another control had no activity. There was no change in the Raven's Matrices scores pre- to posttest intervention, neither were there any between-group differences either pre- or posttest. The arithmetic scores were significantly higher over-all (p < .05) on the posttest, but there were no reliable differences between groups either pre- or posttest. These results suggest that short duration (6 min.) aerobic exercise has no effect on cognitive performance. This finding supports the majority of previous studies that used step-up tasks to examine the relation between physical exercise and cognitive performance.