Abstract: Since the common root-mean-square method and noise power spectrum method cannot eliminate the noise interference recorded by different types of sensors, we use the maximum probability peak displacement as the background noise evaluation index to calculate the noise level more accurate. Based on the reliable noise data, referring to the magnitude maximum distance monitoring capability method and considering the timeliness of earthquake early warning （EEW）, an EEW minimum magnitude evaluation method is proposed. This new method can systematically evaluate the minimum magnitude of EEW and the warning time of independent three types of sensor networks and their fusion networks in Fujian. The results show that the minimum magnitude of EEW based on the fusion networks of the seismometer and the strong seismometer is higher than that of the single seismometer network, and obviously lower than that of the strong seismometer network. For fusion networks of strong seismometer and intensity meter, the minimum magnitude of EEW is similar to that of single intensity meter network. After integrating three types of networks, the minimum magnitude of EEW in 95% regions is about M_L3.2. Because the density of intensity meter network is higher than the seismometer and strong seismometer network, the warning time is the shortest. Compared with the single seismometer network or single strong seismometer network, the warning time of EEW has been significantly reduced by three types sensor fusion networks, and the warning time in 95% regions is 4−6 seconds after an earthquake occurred as it is estimated. This study provides a reference to optimize the distribution of networks and increase the monitoring capacity of key regions in Fujian province.