Predicting the incremental recovery factor from an enhanced oil recovery (EOR) technique is a very crucial task that requires huge investment and expert knowledge to guide EOR laboratory experiments and reservoir simulation studies. Water-Alternating-Gas (WAG) injection is one of the proven EOR technologies. However, WAG injection is a complex process that involves injecting both water and gas under miscible or immiscible conditions. Predictive tools based on machine learning are gaining in popularity in the E&P industry by enhancing conventional procedures and reducing operational cost under current unstable oil market. The aim of this paper is using an ensemble machine learning algorithm to develop a WAG incremental recovery factor predictive model that can be used by reservoir engineers to estimate WAG incremental recovery factor prior kick-off of laboratory experiments and comprehensive technical studies. WAG database preparation followed by ensemble machine learning literature review was performed at first. The used database consists of WAG data from 28 WAG pilot projects worldwide. The database includes rock properties, fluid properties, and WAG injection scheme. The eight input vectors used in this study were rock type, WAG process type (miscible, immiscible), reservoir permeability, oil gravity, oil viscosity, reservoir temperature, reservoir pressure, and hydrocarbon pore volume of gas injected. Based on literature review, Random Forest (RF) learning method was selected to predict the WAG incremental recovery factor and rank the input vector based on their importance. RF develops multiple decision trees based on the random selection of the input data and random selection of the variables. A Random Forest prediction model that predicts the WAG incremental recovery factor as a function of the eight input vectors was developed. The optimum developed model consists of 60 decision trees with a maximum depth of seven (7). The highest important variable in WAG incremental recovery factor prediction is the hydrocarbon pore volume injected. © 2019 SERSC.