Abstract: It is very important to detect shock positions for a shock-fitting technique. However, it is difficult to get the exact position from the computational results. Numerical errors caused by captured methods usually lead to deflections of shock wave positions. In a steady flow, the deflected errors will be eliminated by iterations. In an unsteady flow, however, the deflected errors may be accumulated and led to wrong results. In the present work, a shock detection method based on the theory of characteristics was used to detect shock positions in shock-fitting techniques. This method is proved to be with good performance both in steady flows and unsteady flows.