Reinforced concrete (RC) structures are prone to spalling damage. Successful detection and quantification of spalling are critical to monitoring the structural safety condition. Point cloud generated by emerging surveying technologies such as photogrammetry, laser scanning and Light Detection and Ranging (LiDAR) has been growingly used for spalling damage detection. However, little is known about how to automatically extract all key properties of RC spalling from point cloud data (PCD). This paper presents a three-step computational framework to semi-automatically detect the spalling and quantify its key properties in RC columns from PCD. Specifically, it first removes noise points, calibrates the coordinate system of the captured PCD, and horizontally slices the PCD into thin layers for the spalling damaged RC columns. Secondly, the points for undamaged and damaged areas are detected by measuring the location of each point on its horizontal surface and comparing it with the boundary line of the intact column. The points for the exposed reinforcement is then detected by combining the use of boundary curve fitting and the consideration of the circular shape of vertical rebars. Thirdly, the spalling’s surface area and lost concrete volume are calculated by linear interpolation. A full-size RC column after seismic testing was selected for illustration. The findings contribute to the body of knowledge in structural inspection by introducing a new computational approach for detecting and measuring RC spalling damage. Such an automated approach may largely reduce human interventions and make damage detection and quantification more efficient for post-disaster impact assessment and large-scale building condition assessment.