Abstract: The reliability of CNC rotary tables, as core functional components of machine tools, is crucial for CNC machine tool development. A parameter estimation method based on the nonlinear least squares optimization algorithm is proposed in this study. Fault data from CNC rotary tables were statistically processed and subjected to model assumptions. Unknown parameters were subsequently estimated using the proposed method. The Kolmogorov-Smirnov (K-S) test was employed to verify that the failure data conforms to a three-parameter Weibull distribution model. Based on this validated model, key reliability metrics—including mean time between failures (MTBF), mean time to repair (MTTR), and availability (A)—were calculated for the series of rotary tables. Comparative analysis of the fitting performance between models established using the maximum likelihood method and the correlation coefficient method confirmed the effectiveness of the nonlinear least squares optimization approach. Finally, structural optimization and six Sigma reliability analysis were conducted on identified vulnerable components of the rotary table using the Ansys Workbench platform. This research provides an effective methodological reference for the reliability evaluation and optimization of CNC machine tools and their functional components.