Previous research on social facilitation (e.g. Harkins, 1987) has made it clear that the potential for evaluation plays a role in producing coaction effects, but has left unclear the source of these effects. Subsequent research (e.g. Szymanski & Harkins, 1987 has shown that the potential for self-evaluation and experimenter evaluation each produce performance greater than that found in a 'no evaluation' control group. However, the combination of these sources produces performance no better than either taken alone. In Experiment 1, we show that the potential for evaluation by a coactor also produces performance better than that found in a 'no evaluation' control group, and that the combination of these sources (coactor plus self ) also produces performance no better than either taken alone. Experiment 2 shows that the combination of the potential for evaluation by the experimenter and a coactor does produce better performance than either source taken alone. These findings suggest that, although adding external sources of evaluation contributes to the potency of the coaction effect, adding the potential for self-evaluation to the potential for evaluation by external sources does not.