Smart phones start to equip with MEMS tri-axis accelerometer (i.e. G-sensor) and tri-axis gyroscope chips in recent years for user interface (UI) and game playing purposes. These two sensors actually compose a complete IMU and might be qualified as an INS to aid the GPS positioning of the phones, i.e. a GPS/INS integrated navigation system can be implemented. This paper explores the idea of using the inertial sensors in iPhone 4 from Apple Inc. to make GPS/INS integration for car navigation. A loosely-coupled integrated navigation algorithm with 15-states Kalman filter was used to fuse the data from the GPS and the MEMS inertial sensors. The results of road tests have shown that the MEMS sensors can bridge the GPS position gaps effectively, and can provide attitude estimation at degree level accuracy. The non-holonomic constraint can improve the navigation performance significantly, including both the position and heading. The attitude accuracy can reach the level of 1.4 degrees for tilt, and 2.0 degrees for heading. During the GPS signal outages (e.g. tunnel cases), the position drifts of the MEMS INS are at the level of 30 meters after 30 seconds, with the non-holonomic constraint. Results of this paper proved that the inertial sensors of iPhone 4 can be used for car navigation purpose. They can provide enhanced positioning capability and decent attitude estimation for various applications.