Accurate assessment of fire severity is essential for evaluating the economic loss and ecological impacts from fire, and forest restoration planning. Light Detection and Ranging (LiDAR) can delineate three-dimensional forest structure at individual tree to forest stand level. Here we proposed a new method (tree crown Profile Area Change, cPAC) to quantify fire severity for individual tree from pre- and post-fire LiDAR data. Individual tree crown was firstly segmented from pre-fire LiDAR derived canopy height model. Fire severity was then calculated as the profile area change between pre- and post-fire height percentile curves delineated from LiDAR point clouds within each tree segment. We applied this c PAC method to assess the severity of the American fire, 2013 in Sierra Nevada forest, California, USA. Comparing with plot level field measurements, the fire severity quantified by our cPAC method was strongly correlated with in-situ basal area changes (R²=0.75), whereas other simple LiDAR-metrics (canopy cover change R²=0.36, tree height change R²=0.37) or the Landsat imagery-derived difference in the Normalized Burn Ratio showed much weaker relationships (R²=0.20). Although the general fire spatial extents were consistent, our cPAC method outperformed Landsat-based and other simple LiDAR-metrics in illustrating the tree biomass loss at finer gradient with higher precision.