This study examined the connection between persistence behavior and beliefs about one's own likely or potential performance. Undergraduates (N=114) in teacher education were asked to write test items for weekly topics in a one-semester educational psychology course, based on information conveyed in lectures and text. Subjects could write as many or as few items per week as they chose for 10 weeks. Improvements in grades would be given for the number of items written relative to other subjects. Subjects were told how many points they had earned and were never told where they stood relative to others. At the start of each week, subjects estimated how many items they would write that week and how sure they were about their estimate (perceived self-efficacy), and how important it was and how likely it was that they would earn a bonus in their grade for item-writing (outcome expectations). Responses were used to classify subjects into high, medium, and low self-efficacy groups. Performance on the item-writing task was compared for the three groups over time. The results showed that those who believed they would put in effort and do the work actually did even more than they anticipated doing, while those who expected to do little actually did even less. The findings suggest that there are three distinct groups of performers who differ greatly in their persistence on a self-regulatory task. It was concluded that self-regulated performance is a legitimate motivationally based phenomenon that can be studied, and that students vary greatly in the degree to which they engage in self-regulated performace. (NB)