This paper focuses on a critical analysis and reanalysis of two earlier multisection validity meta-analyses of student ratings of instruction. The reanalysis is based on a sharpening of criteria for: (1) including studies in the meta-analysis; (2) defining key study characteristics that influence results; and (3) calculating the correlational effect size (between instructor rating and student achievement). A subset of 41 well-designed studies and a further subset of 25 high-quality studies were identified, and correlational effect sizes were found comparable to those for the entire sample (for well-designed studies) or slightly smaller (for high-quality studies). The results are seen as supporting the validity of student ratings, especially global ratings of the instructor, and certain rated dimensions such as skill, rapport, structure, interaction, evaluation, and students' self-rating of their learning. Despite overall support, a number of studies showed negative or near-zero effect sizes, but methodological problems may have played a role in producing negative correlations in two studies. (LPG)