Aiming at the problems that the error of inertial navigation system accumulates over time and the Ultrawideband (UWB) positioning suffers from the coarse errors caused by nonlineofsight (NLOS), multipath effect and human body influence, a UWBassisted pedestrian indoor inertial positioning method based on the faulttolerant decision tree is proposed. In this method, a highprecision segmentation fitting error compensation model of the gyroscope is proposed and adopted to suppress the error drift of the inertial navigation; At the same time, based on the UWB assisted pedestrian indoor inertial positioning, a faulttolerant decision tree decision model based on the interaction of inertial navigation and UWB single point positioning data is established, and the coarse error factor of UWB positioning is removed, and then the extended Kalman filter is applied to the parameters of the inertial navigation and UWB, thus the enhanced UWBassisted inertial positioning can be realized. The experimental verification shows that in the complex narrow roadway environment, the method increases the ratio of the mean square error to the length of the route from 6.02% to 0.76%; in the normal square indoor environment, the method increases the ratio of the maximum error to the length of the route from 2.207% to 0.635%. It realizes longterm continuous and reliable positioning and has important value in the engineering applications.