The atmospheric carbon dioxide concentration, [CO2], is a climatic factor that can directly influence future plant growth. The predicted increase in plant biomass in response to increased [CO2] is referred to as the CO2 fertilization effect. Recent results from free-air CO2 enrichment (FACE) experiments indicate that the actual CO2 fertilization effect is significantly lower than expected based on theoretical considerations particularly in crop plants; this phenomenon is called down-regulation and its mechanisms remain still unclear. In this study, we combined statistical modeling of the down-regulation effect with process-based modeling of soybean growth, and integrated the down-regulation effect observed in FACE experiments into a model. As the result of model selection, a model in which the down-regulation only affects the maximum rate of carboxylation was selected. We estimated the future soybean yield under climate change scenarios derived from the MIROC-ESM model. The yields simulated by the model with down-regulation were 15 to 28% lower than those in the model without down-regulation at Illinois FACE site. Moreover, the degrees of overestimation were largely different among areas, which related to the average yield at the areas. These results suggest the importance of integrating the down-regulation effect in even simple models to predict future crop production.