Abstract: The principle of a gate centroid tracking algorithm is introduced. Its tracking error model was established. The error of the gate centroid tracking algorithm in infrared imaging was discussed and analyzed. The centroid tracking error decreases when there is less misjudgment probability, more target pixels and fewer background pixels in the gate. The correct match coefficient increases when there is a greater SNR, fewer correlated searches and more pixels in correlated calculation. Test results show that the algorithm can reduce tracking error, improve tracking accuracy and stability.