Recent results achieved with relative GPS positioning techniques indicate that accuracies at the meter level are possible in land vehicle mode if the cycle slip problem can be minimized. One of the possible solutions to the problem is the integration of GPS and inertial data. Results from inertial surveys show that, with regular, accurate coordinate or range updates, an INS will give velocity estimates at the cm/s level [1]. By integrating differential GPS measurements with an INS, the effect of cycle slips over short intervals may be eliminated from the positioning results. A Kalman filter-smoother to handle this problem has been developed. It integrates differential range and phase measurements with data from an inertial navigation system. The optimal backward smoother improves the filter estimates for periods of poor geometry and multiple cycle slips. The package has been tested with GPS and INS data from a baseline survey near Calgary, Canada. Results of the test show that sub-meter kinematic positioning accuracies and cm/s velocity accuracies are achievable with an integrated GPS-INS.