Abstract: Wildfires can be beneficial for native vegetation. However, wildfires can impact property values, human safety, and ecosystem function. Resource managers require safe, easy to use, timely, and cost-effective methods for quantifying wildfire damage and regeneration. In this work, we demonstrate an approach using an unmanned aerial system (UAS) equipped with a MicaSense RedEdge multispectral sensor to classify and estimate wildfire damage in a coastal marsh. We collected approximately 7.2 km2 of five-band multispectral imagery after a wildfire event in February 2016, which was used to create a photogrammetry-based digital surface model (DSM) and orthomosaic for object-based classification analysis. Airborne light detection and ranging data were used to validate the accuracy of the DSM. Four-band airborne imagery from pre- and post-fire were used to estimate pre-fire health, post-fire damage, and track the vegetation recovery process. Immediate and long-term post-fire classifications, area, and volume of burned regions were produced to track the revegetation progress. The UAS-based classification produced from normalized difference vegetation index and DSM was compared to the Landsat-based Burned Area Reflectance Classification. Experimental results show the potential of using UAS and the presented approach compared to satellite-based mapping in terms of classification accuracies, turnaround time, and spatial and temporal resolutions.