This study determined the effect of exercise on measures of static and dynamic balance used in the assessment of sports-related concussion (SRC). A balanced three-group cross-over randomized design was used with three levels of exercise verified by blood-lactate, heart rate and “perceived-exertion”: no exercise/rest (NE), moderate-intensity exercise (ME), and high-intensity exercise (HE). Participants performed two timed balance tasks: tandem gait (TG) and single-leg stance (SLS); pre- and post-exercise and 15 min after exercise. Linear mixed-models with adjusted means and contrasts compared exercise effects. Ninety asymptomatic participants (45♂:45♀) were recruited. When times were contrasted with NE; HE resulted in a significant decrease in SLS (P<0.001) and TG (P<0.001) performance immediately following exercise. Fifteen minutes of recovery improved SLS (P<0.001) and TG (P=0.011) from post-exercise performance. ME caused a significant decrease in performance in SLS (P=0.038) but not TG (P=0.428). No statistically significant change occurred following ME in any tasks after 15-min recovery (SLS P=0.064; TG P=0.495). Test–retest reliability was considerably higher for the dynamic task compared with the static task. The reliability of static and dynamic balance tasks, and the change in performance following exercise, have implications for the immediate assessment of SRC, as these measures are utilized in concussion assessment instruments.