ABSTRACTThe concept of GPS semikinematic positioning is reviewed, with an emphasis on accuracies at the centimeter level. Details of the mathematical model used herein are given. A two‐step approach combines a Kalman filter to process kinematic carrier phase, phase rate, and pseudorange data, while a batch least‐squares adjustment processes static carrier phase data. A semikinematic test performed over a well‐controlled traverse in a mountainous region near Calgary is used to assess the performance of the algorithm. Several runs of the traverse were made with the station occupation time at 2–3 min. The analysis shows that centimeter‐level accuracies can be achieved when five satellites are observed in good geometery, even when frequent cycle slips occur. An accuracy improvement technique of reverse processing is also investigated. Recommendations for improving semi‐kinematic surveyi