Revealing the relations among robotic comprehensive performance, configuration, scales and working tasks is the basis to optimize robotic mechanism. Due to the correlation and diversity of the single performance indexes, statistical principles of linear dimension reduction and nonlinear dimension reduction are introduced into comprehensive performance analysis and evaluation for typical serial robot. The robotic mechanism ’ s configuration, scales and task with the best comprehensive performance can be obtained by principal component analysis (PCA) and kernel principal component analysis (KPCA) respectively. The results show that KPCA can reveal the nonlinear relations among different single performance indexes more effectively and provide more comprehensive performance information than PCA. Thus, task-oriented method of serial robot for mechanism analysis and evaluation is proposed, which also provides scientific research basis for the mechanism synthesis and optimum task order.