A stochastic model updating method seeking for the probabilistic properties of uncertain parameters is proposed that works better with real-world cases. For simplification, the stochastic model updating process is decomposed into a series of deterministic ones. The Monte Carlo simulation is employed to generate response samples, then combined with the fast-computation feature of the response surface method. An inverse optimization problem is established for predicting parameters corresponding to each sample. Then, the mean and variance of each parameter can be statistically estimated based on the numerous sample predictions. The proposed method has been validated using a set of tested metal plates, and the means and variances of the thicknesses and material parameters are well identified. The proposed method has been validated in order to demonstrate its feasibility and reliability.