In the era of big data, comparing plea leniency and plea bargaining systems can bring important insights to legal practitioners. In this paper, by studying the principle of the fuzzy K-mean clustering method, optimization using rough set theory is proposed for the problem of unclear boundaries in the fuzzy clustering process. Then, based on comparing the origins and development of the plea-leniency system and the plea-bargain system, big data analysis of the plea-leniency system cases in China and the plea-bargain cases in the United States is conducted using the fuzzy clustering method based on rough sets. The application rate of China’s plea leniency system is, on average, 49.67% lower than that of the U.S. plea bargaining system. In terms of the sentencing range, the plea leniency system is 32.73% lower than the plea bargaining system. Compared to the two, the plea leniency system is more stringent in terms of discretionary limits and standards of proof. Based on the big data analysis, the plea leniency system should expand the application area and strict leniency standards and improve the relevant subject system and evidence disclosure system.