Since China claimed to achieve carbon emission peak around 2030 in the “China-U.S. Joint Presidential Statement on Climate Change,” whether or not the target can be accomplished has become the focus of discussion. Thus, the aim of this study is to forecast the carbon emissions peak of Hebei Province in China (as a case study) for the period of 2016-2030 through the historical data of 1990-2015 using the STIRPAT model and GA-BP (BP neural network based on genetic algorithm) model. We choose the proportion of coal consumption, population, urbanization rate, energy intensity, per capita GDP (replaced by GDP in the GA-BP model) and the proportion of services as the independent variables, and set 9 scenarios in the light of different increment speeds of these variables during 2016-2030. Results show that the ranges of estimated carbon emission peaks are 784.1635-1,007.2901 million tons in the STIRPAT model and 702.7465- 702.8144 million tons in the GA-BP model, with corresponding peak years all in or before 2030. Moreover, a comparative study of the STIRPAT and GA-BP models reveals that the GA-BP model estimates carbon emissions more accurately than STIRPAT; however, the STIRPAT model is more precise on the prediction of carbon emission peak years.