Offshore gas reservoirs are characterized by thin interlayers, high production, few wells, etc., and are often exploited by multilayer combined mining, whereas the production dynamics of multilayer gas reservoirs are very different from those of single-layer gas reservoirs. Therefore, clarifying the gas production contribution of each layer in multilayer combined gas reservoirs is an important prerequisite for analyzing the potential of gas reservoirs and realizing efficient development. In this paper, unlike the past method of evaluating the gas production contribution of each layer by using the KH attribute of the reservoir, we combined the modified B-L equation considering CO2 dissolution and the multilayer multizone seepage equation to establish a dynamic split model of the production dynamics of multilayer water-driven gas reservoirs, verified the reliability of the model through the numerical model and the results of the production well logging, quantitatively analyzed the degree of influence of each parameter on the contribution of the layered gas production, and designed the orthogonal experiments. The main controlling factors of the gas production contribution of each layer were determined. The results of the study show that (1) the main controlling factors for the gas production contribution of each layer in the early stage of WDG are, in order, permeability, thickness, outer boundary distance, porosity, CO2 content, and total gas production rate; however, the main controlling factors for the gas production contribution of each layer in the late stage of WDG are, in order, thickness, permeability, outer boundary distance, porosity, CO2 content, and total gas production rate; and the combined view shows that the permeability and thickness have the greatest influence. (2) In multilayer production, the conditions of high permeability, close gas-water boundary, poor gas content, and low CO2 content will reduce the gas production contribution of the layer with the increase of production time. (3) Compared with the results of production logging and numerical simulation, the split model can better predict the gas production of each layer, and the prediction error is no more than 10%. (4) By comparing with the numerical simulation results, the model can realize the prediction of the time of seeing water in the layer with stronger water body capability. (5) The model takes into account the effect of the CO2 content, better reflects the actual gas composition of each layer, and can improve the production prediction accuracy by up to 4%. Considering the high cost of production logging in offshore oil and gas fields, the inability of the KH method to reflect the dynamic changes of gas production in each layer, the poor application of stratified sampling to dry gas reservoirs, and other limitations, the model in this paper can be utilized to simulate the multilayer water-driven gas drive process when the energy of the water body is strong by using the geological parameters of the reservoir and the fluid parameters, and the simulation results of this model provide directions for offshore multilayer water-driven gas reservoirs to improve the recovery rate, and for plugging and regulating the water and exploiting the potential of gas wells that have seen water.