Little is known regarding how attention to emotional stimuli is affected during simultaneously performed exercise. Attentional biases to emotional face stimuli were assessed in 34 college students (17 women) using the dot-probe task during counterbalanced conditions of moderate- (heart rate at 45% peak oxygen consumption) and high-intensity exercise (heart rate at 80% peak oxygen consumption) compared with seated rest. The dot-probe task consisted of 1 emotional face (pleasant or unpleasant) paired with a neutral face for 1,000 ms; 256 trials (128 trials for each valence) were presented during each condition. Each condition lasted approximately 10 min. Participants were instructed to perform each trial of the dot-probe task as quickly and accurately as possible during the exercise and rest conditions. During moderate-intensity exercise, participants exhibited significantly greater attentional bias scores to pleasant compared with unpleasant faces (p < .01), whereas attentional bias scores to emotional faces did not differ at rest or during high-intensity exercise (p > .05). In addition, the attentional bias to unpleasant faces was significantly reduced during moderate-intensity exercise compared with that during rest (p < .05). These results provide behavioral evidence that during exercise at a moderate intensity, there is a shift in attention allocation toward pleasant emotional stimuli and away from unpleasant emotional stimuli. Future work is needed to determine whether acute exercise may be an effective treatment approach to reduce negative bias or enhance positive bias in individuals diagnosed with mood or anxiety disorders, or whether attentional bias during exercise predicts adherence to exercise.