Abstract. This article illustrates an improvement of the traditional MRP algorithm, providing significant benefits in terms of reducing the nervous behavior, better pacing order releases, lowering inventory and concurrently reducing the need for continuous forecast adjustments. The proposed algorithm uses the same set of MRP formulas but refers to a L4L continuous replenishment / production routine, anyway succeeding in releasing batch orders according to EOQ and/or POQ criteria. Simulations have been performed on a three-echelon supply chain in order to test the algorithm effectiveness, either in case of stable but stochastic demand or in case of sudden surges.