This paper presents a method for an indoor pedestrian localization, based on the data that solely are measured by a foot-mounted Inertial Measurement Unit (IMU). To locate the user accurately, a comprehensive Extended Kalman Filter (EKF) with five states is developed. Five different error reduction methods are employed to estimate the errors of all five states. These error reduction methods feed EKF independently, at stance phases or different time intervals of swing phases. The navigation system is developed using the accelerometer and gyroscope measurements and without magnetometer, thus it is insensitive to the presence of metal and magnetic fields, and it is able to estimate the user’s tracked trajectory with the same accuracy in both indoor and outdoor environments.