After reviewing the literature linking threat, incentive, and relaxation to motor performance, Neiss (1988) concluded that both the construct of arousal and the hypothesis of an inverted-U relationship between performance and arousal should be abandoned. These arguments were, however, based on research that does not permit clear evaluation of either the construct of arousal or the inverted-U hypothesis. Furthermore, some of Neiss's assumptions are questionable. This article reviews evidence that provides strong support for the inverted-U hypothesis. It is concluded that the concept of arousal is pragmatically useful in organizing a broad range of behavioral data.