A class of iterative aggregation/disaggregation methods for computation of some characteristics of Markov chain such as stationary probability vectors and mean first passage times is presented and convergence properties of the algoritms are analyzed.