Two theories, the closed loop model (divides self-regulation into self-monitoring, self-evaluation, and self-reinforcement) and the non-mediational model (defines self-regulation as behavior that is controlled by its long-term and observable consequences), have been proposed to explain why behavior changes when self-monitoring occurs. Both theories present empirical predictions which can be subjected to investigation. Based on a matching of their scores on Rosenberg's Self Esteem Scale, undergraduates (N=148) were assigned to either a self-monitor success, self-monitor failure, or a no self-monitoring control group. Subjects completed the Semantic Differential Measure of Emotional States before and after solving sets of five anagrams. Between sets of anagrams, control subjects rested, success-monitoring subjects listed correct solutions, and failure-monitoring subjects listed incorrect solution attempts. Persistence and performance were also measured. Results showed that performance differences among the conditions were nonexistent. Data provided support for the closed loop model of self-monitoring: subjects monitoring failures reported greater decreases in pleasure and less willingness to return for a similar experience than other subjects. These effects were consistent with greater negative self-reinforcement. An unexpected finding was that a substantial number of subjects spontaneously self-monitored their successes. The results provide evidence for the existence of effects that are most consistent with a mediational model of self-regulation. (NRB)