Terrestrial laser scanning (TLS) can acquire high-precision point cloud data within a short time span and has received a lot of attention in the research on forest resource inventory. At present, most of the methods for estimating tree parameters are based on the point cloud of the single tree. This study obtains forest parameters based on the plot-level TLS data. Larch trees in the plot are detected and tree height, diameter at breast height (DBH) and crown projection area of single larch are estimated. The results show that the trees detection accuracy of each plot is relatively high in this study, DBH and tree height can be estimated with relatively higher accuracies with $\mathrm{R}^{2}$ values of 0.949 and 0.77, respectively, and root mean squared error (RMSE) value of2.98 cm and 1.5897 m, respectively. Our results also show that the estimation accuracies of the forest parameters based on the plot-level are similar to that based on the single-level. The proposed method in this study performs much better than the conventional single-level methods in workload and automation.