Prospect Theory (PT) describes the rules of individual's decision-making under uncertainty, which can well characterize decision-makers' risk preference. PT has certain valuable feasibility and applicability in operational decision-making modeling. Based on PT, a method of CGF Agent decision-making modeling was proposed. The Computer Generated Forces (CGF) Agent framework was presented. The detailed description of concrete process and algorithm for course of action (COA) selection in Agent decision-making was also given. The experiment shows that the proposed method has good performance and can improve the authenticity of CGF Agent's decision-making behavior.