Abstract Based on the finite-fault model, combined with the empirical relationship or semiempirical relationship between the moment magnitude and the global source parameters (GSP) and the local source parameters (LSP), the Hybrid Source Model (HSM) of the Yangbi earthquake has been predicted. Considering the regional seismotectonic, crustal structure, seismicity, and semiempirical relationships, the GSP (fault size, average slip, etc.) used in the simulation are given. The LSP primarily includes two parts, one is the asperity parameters describing the deterministic slip, and the other is the k2 model describing the random slip. LSP is determined based on the empirical or semiempirical relationship, and the average value and standard deviation of the GSP are calculated according to the empirical relationship. To generate a series of source parameters that meet the mean and standard deviation, an improved truncated normal distribution function is used. The pseudospectral acceleration (PSA; damp = 5%) of four stations satisfying different geological conditions and orientations are simulated by the stochastic finite-fault approach. The group with the smallest residual error with the average PSA is selected as the final selected focal parameters using the principle of minimum residual error. Eventually, the reliability of this method is verified by comparing it with the inverted source model, and it can be concluded that this method can quickly predict the source model of a given magnitude.