A recent review of the facial feedback literature by Laird (1984) suggested that the effect of facial movement on self-reported mood is large and consistent. In this article, two issues are discussed that suggest that these conclusions are unwarranted. First, methodological problems concerning the facial expressions used to represent valid analogs of emotion and the arousal value of the emotion-eliciting stimuli seriously bring into question the adequacy of the studies to test facial feedback as implied by Izard (1971, 1977) or Tomkins (1962, 1963). Second, even if one accepts the studies designed to represent tests of the effect of facial behavior on self-reported mood, Laird's (1984) box-score approach cannot provide an estimate of the magnitude of the effect. Using meta-analytic techniques (Hunter, Schmidt, & Jackson, 1983; Rosenthal, 1984), I show that the effect size of facial behavior on self-reported mood is actually only of small to moderate value and is most likely an inflated estimate. I conclude, on the basis of the evidence presently available, that the effect of facial feedback on emotional experience is less than convincing.