Abstract: Using the Wangqing forestry area in Jilin Province as a study area, the potential ability of GLAS waveform for estimating forest canopy density was explored in this study. The GLAS waveform was smoothed and fitted by Gaussian low pass filters, and then waveform parameters(i.e. I and e_c)were extracted from the smoothed GLAS waveform. The single-variable model and multi-variable model were developed with the two waveform parameters, respectively. The results showed that the single-variable model built with I were superior to that developed with e_c. The accuracy of multi-variable models was better than that of the single-variable models. For broad-leaf forest, the R²_adj and RMSE(σ_RMSE)were 0.72 and 0.07, respectively, and the validation results were R²_adj=0.74 and σ_RMSE=0.06. The model accuracy of coniferous forest was the highest(R²_adj=0.80, σ_RMSE=0.10)with validation results of R²_adj=0.76 and σ_RMSE=0.11. The model accuracy of the mixed forest was higher than the broad-leaf forest, but lower than the coniferous forest. The values of R²_adj and RMSE were 0.75 and 0.09, respectively, with the validation results of R²_adj=0.71 and σ_RMSE=0.07. The results demonstrate that GLAS waveforms have the capability to estimate the forest canopy density, and the estimation accuracy can be improved by combining I and e_c.