Gas flooding processes have emerged as attractive enhanced oil recovery (EOR) methods over the last few decades. Among different gas flooding processes, CO2 flooding is recognized as being most efficient for displacing oil through miscible displacement. Minimum miscibility pressure (MMP) is a crucial parameter for successfully designing CO2 flooding, which is traditionally measured through time-consuming, expensive, and cumbersome experiments. In the present study, a new reliable model based on feed-forward artificial neural networks was presented to predict both pure and impure CO2-crude oil MMP. Among various properties and parameters, reservoir temperature, reservoir oil composition, and injected gas composition were selected as the input parameters of the proposed model. To evaluate and compare the results of the developed model with existing models, both statistical and graphical error analyses were simultaneously employed. The results showed that the proposed model is more reliable and accurate compared to existing models in a wide range of thermodynamic and process conditions. Furthermore, by employing the relevancy factor, it was found that the reservoir temperature has the most significant impact on the MMP. Finally, in order to identify probable outliers and the applicability domain of the proposed model, the leverage approach was performed. The results illustrated that only two experimental MMP data points were located outside of the applicability domain of the proposed model. As a result, the developed model is statistically reliable for predicting crude oil MMP.