High-density (approximately 40 points ) airborne light detection and ranging (LiDAR) data were exploited to improve the classification of coniferous and deciduous trees in uneven aged mixed wood forests. The investigation was conducted at the individual tree level. Several features were extracted from airborne laser scanning data to characterize structural properties of individual trees, such as crown shape and foliage distribution. A decision tree algorithm was used to perform the selection of significant features and construction of a classifier. The classification was tested within various sites in Canadian forests with different species. The results demonstrated that the LiDAR features describing the foliage distribution within a tree play a significant role in discriminating coniferous and deciduous species. In the classification, 193 reference trees sampled from the entire study area were used for training, and three representative forest sites within the study area were used for assessment of accuracy. The classification accuracy was 77.3%. This study also suggested that high-density LiDAR data were effective in discriminating individual mature coniferous and deciduous trees.