The paper is focused on the search of the optimal values of the planned lead times for the MRP method under lead time uncertainty. The problem is to find the planned lead times, which minimize the expected backlogging and holding costs. The aim of the paper is to give a mathematical formulation of this single-level, multi-item, dynamic multi-period planning problem. For the case where lead time does not depend on the lot size and demand level is constant, an auxiliary Markov model is proposed. Some new formulations of optimal control policies are obtained using this Markov model.