How can we strike a balance between motivating individuals to perform their best without unduly stressing them? In other words, how can we shift our experiences of stress from bad (distress) to good (eustress; Selye, 1956)? One approach is to optimize, rather than simply reduce, the amount of arousal one experiences while performing. Arousal subsumes psychological states of pressure, stress, and anxiety. Using the inverted-U model of arousal and performance as a guide, whereby moderate arousal facilitates performance more so than low or high arousal, this dissertation examines how features of the performance context (including the severity, relevancy, and controllability of a stressor, the uncertainty while performing, and the motivation to perform) can in turn influence cognitive task engagement (task-directed attention and effort). As I show, these contextual and cognitive factors can inform 1) when and how heightened states of psychological arousal may optimize or threaten individuals’ cognitive performance, and 2) the utility and efficacy of emotion regulation strategies to help downregulate arousal. Importantly, across four studies, I show that there is no uniform effect of context, cognitive engagement, or emotion regulation on performance in heightened arousal states. Rather, each factor interacts to determine whether arousal threatens or optimizes performance. Specifically, in Chapters 2 and 3, I examined the arousal-performance relationship by manipulating a performance pressure prior to undergraduates’ completing two cognitive tasks in the laboratory with some (Chapter 2) and no certainty (Chapter 3) in their performance. I show that this transient, task-relevant, low severity, and controllable stressor facilitated performance on a working memory task by increasing participants’ task-directed effort, suggesting that the pressure induced a moderate, optimal amount of arousal. For those in the pressure condition, I also manipulated instructions to reappraise feelings of arousal, but found no difference between reappraisal and no reappraisal groups in any performance or affective outcomes. Because arousal was already optimized, there likely was no need for downregulation of arousal, hence rendering reappraisal unnecessary. In Chapters 4 and 5, I examined the arousal-performance relationship during learning versus performance and with a different stressor: distress about the COVID-19 pandemic. In contrast to the experimental pressure induction paradigm in Chapters 2 and 3, the COVID-19 pandemic is an enduring, task-irrelevant, higher severity, and uncontrollable stressor. I show that higher distress threatened performance by decreasing task-directed attention. Furthermore, in Chapter 5, I show that mindfulness instructions protected individuals from becoming too distracted, though this did not necessarily translate to gains in learning. Thus, these findings suggest that COVID-19 distress pushed individuals to the rightmost side of the inverted-U, where heightened arousal was overwhelming and no longer adaptive. In contrast to Chapters 2 and 3, where pressure optimized performance, here I show that emotion regulation strategies have greater potential utility when arousal is experienced in excess. Thus, in order to optimize arousal and performance, this dissertation highlights the importance of simultaneously considering performance context, its consequences for cognitive engagement, and how these in turn influence the utility and efficacy of emotion regulation.