The generalized Weng model is widely used in oil production prediction. At present,there are linear iterative trial and error method, regression method and genetic algorithm, etc. Linear iterative trial and error method and regression method have uncertainty influenced by artificial factors, which likely cause bigger errors. Genetic algorithm is more complex. In this paper, we solve the generalized Weng model parameter by making use of the nonlinear curve fitting and Levenberg-Marquard iterative method, and establish iterative initial value by the method of multiple regression. Through an example, this method effectively enhances the precision of parameter calculation, and is simple in calculation. It is suitable for parameters calculation of all kinds of time quantum oil and gas reservoir.