Abstract: In evaluation of robotic kinematic and dynamic dexterity performance, there are many different indexes to express robotic kinematic and dynamic dexterity, the global performance index is one of the important evaluation index, including acceleration, angular acceleration and linear acceleration performance index. Those different indexes tend to have different degrees of correlations among them.Some correlations are very remarkable, therefore, the provided information may be overlapped. This paper describes comprehensive evaluation of global performance of the PUMA560 robot with different scales by using principal component analysis (PCA) and kernel principal component analysis (KPCA) , which is characterized by linear dimension reduction and nonlinear dimension reduction principle, then the best robotic scales with comprehensive global performance can be selected.Resultsshow that KPCA method has more effective reduction effects, and can reveal the nonlinear relationship among different single performance indexes to provide more comprehensive global performance evaluation information, which can reveal the numerical calculation retationships among comprehensive global performance and scales, and provide scientific reference for selection of the best robotic scales based on comprehensive global performance indexes.