Delivery cost accounts for a great share in biomass collection cost from fields to a biomass power plant. The main objective of this study was to develop a new methodology to find the optimum geographic distribution of a power plant and its corresponding satellite storages, which was deemed as an effective way to reduce delivery cost under China's specific delivery modes. This study developed a mathematical model for this delivery mode with reasonable simplification and assumption, to explore collection cost composition and establish theoretical function between variable collection cost and the number of satellite storages, in order to obtain the optimum number of satellite storages with minimum collection cost. For spatial analysis of the feedstock supply logistics, a GIS model was developed based on conclusions drawn from the mathematical model. The GIS model can not only give the optimum number of satellite storages, but also the optimum geographic distribution of the satellite storages, their sub-collection-region and the optimum location of the power plant. In the case of a 100 MWe biomass power plant located in Zhanjiang city, in southern China, the site selection models developed were applied to analyze the logistics pathway via satellite storages. The total collection cost can further be decreased by 1% and 9% relative to plant logistics design and direct delivery mode, respectively. Optimization results by the two models, mathematical and GIS, were mutually compared and verified. It was concluded that combining of both models can be highly accurate and efficient to identify the optimum logistics pathway for biomass feedstock from the field to the power plant under specific delivery modes in China. (C) 2011 Elsevier Ltd. All rights reserved.