When Kalman filter is applied to target tracking, the outliers in dynamic measurement data are one of the important factors which affect filtering performance. The reasons and characteristics of outliers emerging in the dynamic measurement data are analyzed, as well as the mechanism that outliers affect Kalman filter. Based on the statistical characteristics of innovation, the method of dynamically identifying outliers and anti-outlier Kalman filter algorithm are presented. The simulation results demonstrate that the method can eliminate outliers and estimate states simultaneously, and it can effectively eliminate the negative effect of the outlier on the filter, and improve the filter accuracy.