Abstract: After the development of the oil field, it is an important task to predict the production and the recoverable reserves opportunely by the production data. At present, the forecasting methods include reservoir numerical simulation, forecasting model, material balance and production decline ones, and so on. Of these methods, the forecasting model is a useful and effective one for entire forecasting. The famous generalized Weng model, Weibull model, Rayleigh model, HCZ model, Hubbert model, Lognormal distribution model are major forecasting models based on the development model of a single peak, by which the reliable results can not be obtained for dual?peak and multi?peak development model. In this paper a multi?peak forecasting model and its nonlinear multi?parameters automatic matching techniques are proposed, which are effective for the multi?peak development model. According to the practical application, this multi?peak forecasting model is practical and effective