M_L,M_WP and m_B are commonly used methods of magnitude measurement in automatic earthquake rapid reports. In order to analyze the accuracy of these three methods at different times and demonstrate the reasons of deviations,we simulated the quasi-real-time determination process of the magnitude of the M6.4 earthquake in Jiashi,Xinjiang(January 19,2020)and the moderate-strong earthquake swarm in Yangbi,Yunnan(May 21,2021). By calculating the M_L,M_WP and m_B  magnitudes of 56 earthquake events with M4.5 and above occurred in mainland China from January 1,2020 to July 31,2022,we analyze the stability of these three methods in the determination of moderate-strong earthquake magnitude,after comparing their own benefits and applicability study in different magnitude section. Our results showed that:① Using the conversion relationship converting M_L to M is an important reason for the large deviation. M_L should not be converted,and its value can be directly used as the published magnitude of the automatic quick reports in moderate and strong earthquake events;② For densely occurring earthquake swarms,M_WP and m_B are easily affected by the previous coda wave or long-period signal,resulting in large magnitude and large deviation,so it is not suitable to be used as the magnitude of the automatic rapid reports,but M_L is better;③ In the 4.5～5.5 magnitude range,M_L should be used to measure the magnitude;in the 5.5～6.5 magnitude range,both M_WP or m_B  are good,but M_WP is more stable and time-saving for the earthquakes. For magnitudes above 6.5,M_WP has the better effect and should be applied.