The outlier detection and accommodation of integration navigation of strapdown inertial navigation systems and global position system (SINS/GPS) were studied. Based on analyzing the innovation orthogonal property in Kalman filter, an outlier adaptive detection approach was first presented, which included the determination of evaluation function and threshold and the logic decision of outlier occurrence. To effectively attenuate the influence on estimation accuracy, a modified Kalman filter algorithm was proposed by accommodation of the dynamic data with outlier. Results of data processing from vehicle-test SINS/GPS integration navigation show the effectiveness of the proposed method.