Abstract: To promote the engineering application of reliability allocation methods in the research and development of CNC machine tools under enterprise practical conditions, a reliability allocation optimization algorithm based on a trade-off improved cost-effectiveness ratio was proposed. Firstly, a reliability prediction method for multi-source data fusion is established, utilizing the conversion relationship between reliability data and the Bayesian fusion method, to complete the initial evaluation of subsystem reliability. Secondly, by combining the analytic hierarchy process (AHP) and the ordered weighted averaging (OWA) algorithm, a relative rating is assigned to enterprise personnel based on actual improvement measures and corresponding relative amounts. Group decision-making is used to scientifically quantify the functional relationship between improvement costs and benefits and the increase in subsystem reliability caused by specific improvement measures, thus establishing a quantitative model for subsystem improvement cost-effectiveness. Thirdly, specific optimization allocation for reliability implementation will be constructed using the non-sorting genetic algorithm (NSGA). Finally, taking the five axis CNC machining center as an example for method validation, the mean time between failures (MTBF) allocation values of each subsystem were determined, revealing weak components such as the spindle system, feed system, and CNC system, while providing quantitative reference for the selection of improvement measures for the key subsystems mentioned above. The paper realizes the reasonable allocation of reliability for CNC machine tools, which has important guiding value for the reliability design of machine tools.