Although measures of perceived message effectiveness are commonly employed in formative evaluation research, the validity of such measures has been assumed rather than demonstrated. A meta-analysis revealed a corrected-for-attenuation correlation of .41 between perceived effectiveness (PE) and attitude (k=40). This association was apparently subject to the influence of moderator variables, although the pattern of influence was unstable across different types of analysis. Overall, the results empirically demonstrate the value of PE judgments in formative research and in circumstances in which actual effectiveness (AE) is difficult or impossible to measure. They also prompt a host of questions concerning (a) the causal relationship between PE and AE, (b) the possibility of bias in these judgments, and (c) the potential for developing methods to reduce or remove bias in effectiveness judgments. Future research is needed to corroborate these results as well as broaden the range of AE measures.