The invention discloses a square root volume Kalman filtering method in a non-linear target tracking system, which aims to an adaptive square cubature Kalman filtering (ASCKF) algorithm, improves the time updating link and the Sage-Husa noise estimator link and forms an ASCKFNS algorithm. The method of the invention is suitable for a non-linear discrete system and further increases the algorithm accuracy on the premise of ensuring the stability of the algorithm. In addition, computation amount is reduced and the real-time performance is increased.