Security-Constrained Optimal Power Flow (SCOPF) for a large-scale power system is a complex and difficult problem with high computational requirements. To solve the problem, this paper proposes a constraints reduction method based on the security operation domain of a power system. The proposed method includes two technologies. First, the monitored branches are determined according to the actual physical meaning and the mathematical relations between different post-contingency constraints, so that the security operation domain can be formed by a group of inequality constraints. Secondly, multi-layer screening, including residual coefficient, transfer coefficient, and mathematical depth filtering are performed on the post-contingency constraints according to the transfer characteristics of the power system, so that the number of inequality constraints can be greatly reduced. Simulation on the IEEE-118 system and an actual power system verifies the practicability, effectiveness, and robustness of the proposed method in solving these engineering problems. This work is supported by the National Key Research and Development Program of China (No. 2017YFB0902800) and the Science and Technology Project of State Grid Jiangxi Electric Power Company Ltd. (No. 521810190001).