Abstract: The goaf boundary affects the mining quality of adjacent stopes during the two-step mining process in underground mines, but complete boundary data cannot be obtained well by traditional measurement methods. In this paper, the unmanned aerial vehicle 3D laser scanning measurement system was used to scan and measure the goaf in an underground iron mine. The maximum single point error of the scanning data was approximately 3.6 cm, and the distance error between any two points was approximately 2.0 cm, indicating its high scanning accuracy. On the basis of that, the 3D model of the adjacent goaf was soon obtained, and the dilution and loss indexes of the stope in the first step were predicted and analyzed. It was concluded that the increase value of the actual room dilution rate due to overcutting room in the first step was 3.46%, while the rate of theoretical room loss caused by pillar loss was 13.11%. The above results provide data support for the actual two-step stope design and lay a foundation for optimizing the stope design scheme to maximize the recovery benefit of the stope.