This study examined the system of wood pellet distribution to residential consumers. The distribution cost for a residential pellet consumer typically represents 30% of the per tonne price and of this share, the inventory cost could be more than 50%. Important administrative activities in physical distribution are forecasting demand and inventory control. One way to improve distribution systems would be to optimise inventory management for pellet distributors. The aim of this study was to propose improvements in pellet distribution management by using tools from systems analysis. The ultimate goal was to present an optimised storage level curve adapted to the mid-Swedish community of Avesta. An internal model for optimising inventory management, Pell-Sim, was constructed, composed of two integrated parts: a simulation unit to forecast residential wood pellet demand and a spreadsheet unit with inventory-related functions. Daily outdoor temperatures basically regulated the simulation unit. An order point system was chosen for reordering. The residential customers of a distribution company were divided into two groups, delivery and collecting customers, which were statistically treated separately. When collecting and delivery customer input inventories were normally distributed in the intervals from 0 to 3500 kg and 6500 kg, respectively, their annual means of total delivery were both about 7000 kg/customer, which was the desired and empirical level. The expected pellet customer orders were negatively correlated to mean daily temperatures, lagging behind about I month. Sensitivity analyses showed that monthly results for ordered quantity and total cost were particularly sensitive to ordering and carrying costs. The Pell-Sim programme can easily be adapted for distributors in other geographical regions. (C) 2004 Elsevier Ltd. All rights reserved.