We consider two time-inhomogeneous discrete Markov chains with close one-step transition probabilities - in the uniform total variation norm. The problem of the stability of the transition probabilities for arbitrary number of steps is investigated. Themain assumption is the uniform mixing. For example, we prove that the difference between distributions of chains after any number of steps not exceeds the value [epsilon]/(1 - p), where [epsilon] is the uniform distance between transition matrixes, and p is the uniform mixing coefficient. Proofs are based on the maximal coupling procedure that maximize the one-step coupling probabilities.