In pedestrian navigation system, zero velocity update (ZUPT) can estimate the velocity errors and horizontal attitude errors. However, good heading error estimation for such a system remains a challenge, this is due to the unobservability of heading error. In order to solve the problem of easily divergent heading error, zero integrated heading rate (ZIHR) correction method is proposed. The heading angle difference at adjacent time has a certain relationship with the drift of the gyro and heading error angle. The difference is taken as the measurement value and one-dimensional measurement is extended on the basis of ZUPT. The errors which are estimated by Kalman filter are fed back to original navigation system. In the end, multiple sets of physical experiment results show that the ZIHR correction method can efficiently reduce the drift errors of the micro-electro- mechanical system inertial sensors, and the position accuracy can reach 2% of travel distance.