Objective To explore the quantitative analysis and evaluation method for crawling abnormalities in children with cerebral palsy ( CP) based on kinetic and surface electromyography ( s EMG) characteristics. Methods Children's hand-knee crawling was studied in this paper. Firstly, the inertial sensors and s EMG electrodes were used to capture the features that effectively described the intensity, stability, muscle activation level and coordination of antagonistic muscle pairs during hand-knee crawling. Secondly, aiming to solve data redundancy problem in the overall quantitative evaluation of crawling motor function from multichannel data features, the principal component analysis ( PCA) algorithm was used for feature fusion, and the quantitative evaluation method of crawling motion abnormality was proposed based on feature distance ratio.Results There were 17 healthy children and 22 children with CP recruited in this study. The experimental results demonstrated that PCA-based feature fusion could reduce the redundant information in multi-dimensional features, and the evaluation method based on feature distance ratio could effectively describe the abnormal crawling motion of CP children. Conclusions This method based on kinematics and s EMG feature distance ratio provides a possibility for quantitative evaluation of crawling abnormalities in children with CP.