The processes by which individuals establish, strive for, and revise personal performance goals were examined in a longitudinal study of track and field athletes. Participants were followed over the course of their competitive season and completed goal-progress questionnaires after each competition. Hypotheses regarding initial goal choice and goal revision were drawn from control theory and social cognitive theory, and comparisons of the two theories were conducted. Results supported social cognitive theory's assertion that self-regulation involves both discrepancy-production and discrepancy-reduction processes. The majority of athletes created positive goal-performance discrepancies by setting initial goals higher than previous best performance. In addition, goal revision during the season served to maintain, rather than eliminate, goal discrepancies. Goal revision was predicted by the size of the discrepancy between one's goal and current performance, moderated by performance attributions and phase of season. Large discrepancies that were attributed to uncontrollable causes or that occurred later in the season resulted in downward goal revision. Goals were increased when discrepancies were small and attributed to controllable causes. Implications of these results for theory and research on self-regulation of goals and performance are discussed. Personal goals are seen as central organizers and regulators of human action in several contemporary theories of human motivation (e.g., Bandura, 1986, 1991; Carver & Scheier, 1981; Dweck Br Leggett, 1988; Emmons, 1991; Locke & Latham, 1990). These theories all portray individuals as using goals to monitor, evaluate, and change behavior. However, there are subtle yet important differences between these theories in terms of the underlying processes responsible for goal-regulated behavior. The purpose of this article is to compare predictions made by two theories of self-regulation: control theory (CT) and social cognitive theory (SCT). Reconciling differences between theories and integrating them (if possible) is important because of the growing popularity of goal setting and continuous-improvement programs in organizations. For these programs to work effectively, it is necessary to understand precisely the mechanisms that govern the dynamic regulation of performance according to goals and standards. Using a sample of track and field athletes, we use CT and SCT to address two specific questions related to dynamic self-regulation: (a) What level of performance goals do individuals set for themselves when beginning a performance cycle, and (b) what mechanisms guide decisions to revise or persist in pursuing goats when faced with feedback that performance is discrepant from goals?.