We present a forecasting model for shale gas production based on different gas flow mechanisms and discrete fracture models. The production forecasting model takes into account the gas rarefaction and stress sensitivity effect and is validated with two field examples. The present model can reliably predict gas production in shale reservoirs. The effects of stress sensitivity, the half-length of hydraulic fractures, the initial conductivity of the hydraulic fractures, and the initial pore diameter on shale gas production are discussed. The results show that the permeability and porosity are overestimated without considering stress sensitivity, which makes higher daily gas production in the early stage of shale gas development. For well 73 (W73) and well 314 (W314), the cumulative productions without stress sensitivity are 10.78 and 4.96% higher than those with stress sensitivity, respectively. The cumulative production increases with increasing the half-length of hydraulic fractures in 1000 days. When the initial conductivity of hydraulic fractures increases, gas transportability in the hydraulic fractures increases. However, the gas supply ability of the shale matrix does not increase with the increased initial conductivity of hydraulic fractures. The half-length of the hydraulic fracture plays a major role in cumulative production. This indicates that increasing the half-length of hydraulic fractures is an efficient way to enhance gas production. This paper provides some theoretical support in gas production forecast and hydraulic fracturing optimization.