The steps of theoretical aspects of univariate, adjusted univariate and multivariate analysis for repeated measures studies are presented. Some formulas of the test procedures are extended to make computation easy. As a verification and to provide background experience of the test procedures, some recent data from livestock, clinical, psychological and physiological experiments are analysed and the results are incorporated with the theoretical discussion. Considering the repeated measurements experiments model several wellknown procedures are presented for multiple inferences of the occasion contrasts. The adjusted and multivariate tests for multiple comparisons of repeated measures means are also considered when the sphericity requirements are not considered valid. A simulation study is made to compare the univariate, adjusted univariate and multivariate methods. The sample variance-covariance matrices observed from a number of experimental studies as representative of population variance-covariance patterns are used to generate sets of multivariate normal 'lq observation. The sphericity and homogeneity assumptions of the matrices are presented and the comparisons are interpreted in the light of these assumptions. Actual (fractional) degrees of freedom of the test statistics are used and an investigation is then presented which compares several data analytic strategies when the assumptions are violated. The question of how to analyse repeated measures data with missing observations is a common problem facing analysts. An attractive approach is to estimate the missing values and then proceed with the standard analysis. A new method for values which are missing at random is discussed and compared with other methods, and the new method found best.