Residuals in the GMANOVA–MANOVA model with rank restrictions on the mean parameters is considered. The main objective is to define residuals useful for evaluating the reduced rank restriction model. We decompose linear spaces into four subspaces as it can be done for the Extended Growth Curve model with two “profiles”. The new residuals are defined by orthogonal projections on these subspaces. It is discussed how the new residuals can be used to test model assumptions.