Abstract: In order to determine the main causes of falling accidents in high building s construction, we put forward a causation set system for falling accidents based on HFACS, and reveal the corresponding relations between accident sample cases and causations. We determined some prerequisite conditions for falling accidents and built a causation framework; then constructed a contingency table between accident samples and some causative factors based on causal relationship between accidents and causes. Applying the principles of CAM, we described relationship between sample points and causes after carrying out multivariate statistical analysis on sample cases and causative factors of a two-dimensional plane. According to the clustering characteristic of sample points, a typical sample point identification function was established using Circular Cover Method (CCM) to ascertain typical sample points; and we further put forward criteria for determining the main causes and conducted an empirical study according to the corresponding relationship between causal points and typical sample points. We chose twenty cases from construction falling accidents report, and they were all verified by CAM. The results show that there are seven factors (e.g. insufficient education and training, poor team management) contributing to falling accidents in building construction.