Forest height is an important indicator for forest biomass. Stereo imagery is sensitive to the vertical structure of forest. However, the result directly derived from stereo imagery is digital elevation model. The ground surface elevation under forest canopy is needed to derive forest height from stereo imagery. Considering the penetration abilities of L-band SAR data, the combination of stereo imagery and the L-band InSAR for the estimation of forest height was proposed. Their difference image was compared with the forest height provided by lidar data. The forest patches exhibited in lidar data were also clearly visible in the difference image. It is feasible to extract forest height information from the combination of PALSAR InSAR data and PRISM stereo images.