Efforts for nuclear non-proliferation have continued since the development of nuclear weapons and the conclusion of the NPT Treaty. Nuclear proliferation requires materials, facilities, and human resources to make nuclear weapons, and it takes a medium to long-term time. There are many restrictions in the current system to obtain nuclear materials and facilities, so it is often done through illegal means, black markets, or confidential transactions. Methods have been developed to evaluate the nuclear non-proliferation regime to strengthen the non-proliferation and solve the problems. The IAEA and the United States DOE initiated the proliferation resistance evaluation in 1980. The DOE conducted the assessment in three main evaluation categories: materials, technical characteristics of facilities, and institutional barriers. In another nuclear non-proliferation evaluation study, some researchers evaluated three main types: current capacity, political situation, and international situation. Detailed indicators include economic capacity, industrial capacity, nuclear capacity, leader’s intentions, political structure, competitive relations, alliances, and international norms. Most of these evaluations are based on the situation at the time of assessment at the national level. Historical examples of nuclear proliferation are rare, and verification is also challenging. The Bayesian probability is widely used when the data is small, experiments are impossible, and the causal relationship is unclear. A Bayesian network is a combination of Bayesian probability and graphics. It is used throughout the industry because it can easily derive results according to causal relationships and weights of various variables, evaluate the risk for decision-making, and obtain changed results through data updates. In particular, to evaluate the proliferation of nuclear weapons, Freeman developed the Freeman network in 2008 and the Freeman-Mella network in 2014. Freeman explained in detail only the process of deriving variables, correlations, and probabilities of factors related to factors such as motivation, intention, and resources. It isn’t easy to view as an objective result value because it does not describe the academic background for path selection, motivation list, intention, and resource variable selection. However, the research was meaningful because he first used the Bayesian network for nuclear proliferation. Although some studies have been done at the macro level, there is no case of applying it in export controls, which is the beginning of the actual spread. Also, there is no quantitative value for factors for risk assessment. There is little data, and verification of causality is difficult, so if the Bayesian network is applied to export control and applied to actual implementation, it will help make decisions such as export license or export denial.