Abstract: Inventory and transportation cost the main expenses of logistics. A new combined optimization policy is researched in this study, on the Inventory Routing Problem (IRP) in a three-echelon distribution system, which composed of a vendor, a Distribution Center (DC) and multiple geographically dispersed retailers. In the system, each retailer faces a deterministic, retailer-different rate and same type of product demand. Any retailers demand must be replenished from the DC, not permitted from the vendor directly. The stocks may be kept at any retailers or the DC. The objective is to determine a combined transportation (routing) and inventory policy to minimize a long horizon average system cost without shortage or delay. After discussing the characteristics of Fixed-Partition and Integral-Ratio Horizon (FP-IRH) method, the model for the IRP of three-echelon system is constructed. Then, a decomposition solution is proved, where the retailers are partitioned into disjoint and collectively exhaustive sets and each partition of retailers is served on a separate route. Given a fixed partition, the original problem is decomposed into two-echelon sub-problems and the efficient algorithms are designed for the three-echelon IRP by developing solution of the famous Vehicle Routing Problem (VRP). The computational results for a testing example problem show that the FP-IRH policy is more effective.