Hypotheses that do not have the standard bilinear form theta=CBM=0 occur naturally in the analysis of repeated measurement designs. An expanded class of the tests of the form psi=Tr(G theta)=0, called extended linear hypotheses, provides a richer class of parametric functions. A method to analyze double multivariate and mixed multivariate models (MMM) using the Statistical Analysis System (SAS) is demonstrated. The analysis is extended to extended hypotheses, and a new approximate test of extended linear hypotheses for MMM designs is developed that does not require multivariate sphericity, but only a general Kronecker structure. Two appendixes provide SAS programs for these methods. (Contains 4 tables and 26 references.) (SLD)