The zerospeed interval detection algorithm in the pedestrian inertial navigation based on MEMS is an important factor restricting the error of navigation position solution error. Aiming at the problem the zerospeed detection algorithm using the fixed threshold has poor detection adaptability under different motion states, a zerospeed detection algorithm based on adaptive threshold is proposed according to the pedestrian gait characteristics and periodic zerospeed trends, taking the resultant acceleration and angular velocity amplitude as the detection data. Based on the dynamic and statistical characteristics of the detected data in the motion state, the zerospeed interval detection threshold is updated according to the obtained motion state information to adapt to achieve the accurate zerospeed interval detection in different motion states. The experiments show that the adaptive threshold algorithm can accurately detect the zerospeed interval, and the accuracy of zerospeed detection reaches over 98%. The error rate of detection results used for the navigation and positioning calculation is less than 1.5%.