The common interpretation of the positive correlation among self-efficacy, personal goals, and performance is questioned. Using self-efficacy theory (A. Bandura. 1977), it was predicted that cross-sectional correlational results were a function of past performance's influence on self-efficacy, and using control theory (W. T. Powers. 1973), it was predicted that self-efficacy could negatively influence subsequent performance. These predictions were supported with 56 undergraduate participants, using a within-person procedure, Personal goals were also positively influenced by self-efficacy and performance but negatively related to subsequent performance. A 2nd study involving 185 undergraduates found that manipulated goal level positively predicted performance and self-efficacy positively predicted performance in the difficult-goal condition. The discussion focuses on conditions likely to affect the sign of the relationship among self-efficacy. goals, and performance.