This research investigated the effects of anchoring biases on judgments of self-efficacy, and the perseverance of initial efficacy judgments as subjects received firsthand performance feedback. Prior to attempting a novel task, subjects judged the level of performance they could achieve in relation to an anchor value representing a low, intermediate, or high outcome. Anchoring strongly affected initial self-efficacy judgments. Subjects then performed the task; success rates were controlled and equivalent in all conditions. On subsequent self-efficacy assessments, the initial anchor-biased differences in perceived self-efficacy were evident despite extensive performance feedback. Initial self-efficacy judgments significantly predicted subsequent level of self-efficacy within the anchoring conditions, but not within a no-anchor control group.