Abstract: Forest stock volume is an important indicator of forest resource survey. Exploring the potential of combining active and passive remote sensing to estimate forest stock volume is of great significance for forest resource monitoring. Xixiaoshan Forest Farm in Linjiang City, Jilin Province is taken as the research area, and Sentinel-1 and Sentinel-2A are used as the data source to extract 80 texture variables, 14 optical characteristic variables and 3 backscattering coefficients. Forest stock volume estimation models are compared based on multiple stepwise regression methods and random forest methods to determine the best method for forest stock volume estimation. Among the single type of characteristic variables, model (a) constructed from texture variables was better with a Root Mean Square Error of leave-one-out cross-validation of 53.91 m³·hm⁻². Among the five forest volume estimation models, Model (d) with multiple types of parameters was better estimated with an R² of 0.513, a Root Mean Square Error of leave-one-out cross-validation of 49.70 m³·hm⁻², and a relative Root Mean Square Error of leave-one-out cross-validation of 0.26. In addition, Xixiaoshan Forest Farm field show a spatial distribution of forest stock volume with low in the south-central regions and higher in the east and central west regions. The fusion of Sentinel-1 and Sentinel-2 data can improve the accuracy of forest storage estimation.