In two experiments, self-efficacy theory was used to conceptualize the effect of efficacy expectancies and valence of self-evaluation on performance. In Experiment 1, efficacy expectancies were manipulated by practice item difficulty and feedback, and self-evaluation was manipulated by providing a standard of comparison. In Experiment 2, efficacy expectancies developed spontaneously while participants performed easy or difficult tasks, and the potential for self-evaluation was crossed with the potential for experimenter evaluation. In both experiments, as predicted, in the high-efficacy condition, self-evaluation participants performed better than no-self-evaluation participants, whereas in the low-efficacy condition, self-evaluation participants performed worse than no-self-evaluation participants. In addition, in Experiment 2, the equivalence of self-evaluation and experimenter evaluation was demonstrated. Implications for self-efficacy and self-evaluation processes are discussed.