The Yellow River basin of Shandong province is the area close to the sea, where is one of the most serious soil erosion areas within the basin. It is very important to optimize the distribution of forest resource for water and soil conservation on the regional scale based on remote sensing images and mathematical methods. In this study, two periods (2000 and 2006) of TM remote sensing images with similar temporal phase were selected as the basic data sources. The supervised classification, expert classification and GPS, GIS techniques were used for classification, statistics, verification and analysis of the forest resources. The results indicated that the forest resources spatial distribution was extremely uneven, with main concentration in the south-central mountainous and hilly area. The growth of the forest resources was relatively slow over the 6 years. The area of the barren hills and lands suitable for afforestation was increased by 8.2% during the period from 2000 to 2006, which amounted to 238955.7 hm². The barren hills and lands suitable for afforestation were optimized using the linear programming method. On the regional scale, the forest types and structure were significantly improved, and land use structural information entropy was reduced by 8.4% after the optimization. The spatial structure order degree of the forest resources and land use were improved. Finally, the corresponding forest management measures for the different site conditions were given.