The information of the foliage cover and the distribution of branches are essential sources for understanding the spatial variability of the vertical forest structure. But it is difficult to use traditional research methods, such as doing field surveys and interpreting aerial photographs, to obtain related information on the vertical forest structure and conditions below the canopy. Therefore, this study attempted to evaluate the possibilities of using airborne LiDAR data to examine the forest vertical structure below the canopy, and utilize LiDAR data to classify land cover types in mountain areas. Red cypress (Chamaecyparis formosensis), Sugi (Cryptomeria japonica), mixed hardwoods, and bare land were the 4 categories analyzed in the investigated area, the Alishan region of central Taiwan. The analytical methods were based on LiDAR multiple-return and intensity data, and statistical analyses and image classification were used to describe the diversities of the investigated land types. The ratio of echo model (REM) and echo intensity model (EIM) were effective in distinguishing the divergences of different land types. Results of this study demonstrated the proportion of echo return and intensity data related to the canopy density. Among the types of test echoes used in the study, plentiful information for land cover classification using both the ratio of echo returns and the intensity of echoes was acquired from the first echo returns. The results of applying singleimage classification showed a classification accuracy of 50.5~68.5%. The EIMFE, REMFE, REMLE, and REMOE showed a higher potential for classifying land types. The classification results of stack images indicated that combining more LiDAR-derived variables yielded a more-accurate classification accuracy (81.5%). This study corroborates the high feasibility for mapping land cover types using LiDAR multiple-return and intensity data.