The book is an attempt to establish systematic overwiev of solutions of all basic statistical problems in multivariate statistical models. As basic problems are considered estimation of parameters of the mean value of the observation matrix of the model, estimation of covariance matrix parameters, construction of confidence regions and testing statistical hypotheses. All problems are solved in multivariate models without constraiants on parameters, in models with the type I constraints and in models with type II constraints. Relatively great attention is devoted to a construction of insensitivity regions, which enable us to judge whether the estimators of covariance matrix parameters can be used in inference on parameters of the mean value of the observation matrix.