Accurate prediction of the force on the cutting tool and workpiece is of great significance for the micro-milling process safety. To improve the machining precision and the sustainability of micro-cutting tools, the uncertainty of parameters and the inaccuracy of measurement in actual manufacturing should be considered. In this work, an improved cutting force model is proposed, in which the tool runout and the size effect are involved, and a new method to calculate the cutting coefficient is presented. Adaptive Bayesian updating based on structural reliability analysis and adaptive-kriging method is developed to solve the actual distribution of parameters with uncertainty. The actual cutting force data is measured by the micro-milling experiment with workpiece material Al6061. The posterior parameter distribution characteristics are obtained and verified based on the experimental data, the proposed Bayesian method, and the established mechanical model. The reliability analysis is used to evaluate the influence of various parameters on the micro-milling system and explore its reasonable selection range. The results show that the calculation efficiency of the proposed Bayesian method is significantly improved, and the conclusion of reliability research can provide helpful guidance for complex practical machining.