Meta-analytic methods were used to summarize results of Monte Carlo (MC) studies investigating the robustness of various statistical procedures for testing within-subjects effects in split-plot repeated measures designs. Through a literature review, accessible MC studies were identified, and characteristics (simulation factors) and outcomes (rates of Type I error) of each MC study were coded for univariate, df-adjusted univariate, and multivariate test procedures. Results of weighted least squares regression indicate that all procedures are generally robust to violations of the multivariate normality assumption. The e-circumflex F-test was generally insensitive to departures from the sphericity assumption, with degree of bias decreasing with increases in the degree of non-sphericity. For balanced designs, all test procedures were reasonably robust to moderate degrees of covariance heterogeneity, with Type I error rates becoming only slightly liberal with increases in the degree of covariance heterogeneity. When the design was unbalanced, however, all procedures were sensitive to the presence of heterogeneous covariance matrixes, particularly for the within-subjects interaction effect, where the tests became increasingly conservative or liberal depending on the pairing of unequal covariance matrices and unequal group sizes. For balanced designs, either a df-adjusted univariate or multivariate approach is recommended. For unbalanced designs, the generalized approximate (GA) or improved GA procedures of H. Huynh (1978) are recommended rather than the investigated test procedures. Eleven tables summarize analyses. (Contains 49 references.) (SLD)