Three representative small and medium earthquakes in Shanxi seismic belt in recent years were taken as examples. By using different methods, such as the first motion of P-wave, first motion of P-wave combining P/SV/SH amplitude ratio, CAP (Cut and Past), and moment tensor inversion, the reliability and stability of focal mechanism solutions of small and medium earthquakes were analyzed and verified by gradually increasing calculation stations and selecting different data types and data quantities. The results indicate that in practice, 1) for a stable and reliable solution based on P-wave first motions, at least more than eight P-wave polarity data with clear initial motion and good inclusion in the earthquake are required; 2) for small earthquakes with inadequate station records, increasing the amplitude ratio data improves constraints on the results to a certain extent, but increasing the amplitude ratio data often significantly increases the contradiction ratio, while the constraints on the results are not continuously improved; 3) for ML≈3.2 earthquakes, the CAP method gives stable and reliable results, and for medium earthquakes with larger magnitude, moment tensor inversion method is also suggested to be an ideal choice to obtain reliable focal mechanism solutions.