Abstract: In order to send out earthquake alarm information timely, accurately and reliably, an early earthquake alarm method using P-wave double-parameter alarm method based on Bayesian theory is proposed in this paper. Based on 355 earthquake data recorded at Sichuan Province of China Earthquake Networks Center, the median period τ_c and ground motion amplitudes P_d in the first 3 seconds after P-wave triggering are calculated. Combined with Bayesian theory, the prediction model of magnitude and peak ground acceleration is established. Taking the magnitude M＝4.5 and PGA＝120 cm/s² as the alarm threshold, the seismic hazard discrimination model is established. Compared with the traditional fitting method, the simulation analysis of earthquake early warming based on Bayesian theory, and then the Wenchuan M_S8.0 earthquake is taken as an example to carry out the earthquake hazard identification. The experimentalresults show that early earthquake alarming method using seismic P-wave double-parameter based on Bayesian theory is 15.15% lower than the traditional fitting method, which can can quickly and accurately estimate the magnitude and peak acceleration, and effectively assess the hazard of earthquake, which can provide data support and decision-making basis for earthquake monitoring and alarming.