Abstract Ranking of best possible Enhanced Oil Recovery (EOR) technics for implementing on a target field is one of the most important questions that should be answered by reservoir engineers. EOR screening can be considered as a tool for recommending the most appropriate EOR methods. Although for each candidate reservoir, the applicability of EOR processes must be investigated specifically, EOR screening can be used as an indicator before economic evaluations or reservoir descriptions are done and executive decisions are made. Implementing an EOR project for predictions that pass this screening is the next step. In this study, the fuzzy decision tree method (with the ability to rank and classify EOR methods simultaneously) is introduced for EOR screening. Basic features for this study are permeability, viscosity, depth, temperature, saturation, and API. Using a fuzzy decision tree enables us to design an expert system which generates EOR rules automatically. This is one of the noticeable features of this study which reduces the importance of a human expert role while designing the system and making it as expert as possible. Here, the fuzzy decision tree method is implemented on a dataset consisting of 548 observations related to 10 different EOR techniques. Predictions made by this method which are ranked from the most applicable EOR method to the least one include the EOR method mentioned in the dataset for every observation in both training and test set. Moreover, using the procedure introduced here for training the trees enables the expert system to be adaptive whenever the dataset is updated. Highlights • FID3 trained in this study, predicts all observed EOR methods with the truth level threshold equal to or greater than 0.7. • This approach suggests a ranked set of appropriate EOR processes for each set of inputs containing the observed EOR method. • EOR selection is done among each group as well as all of EOR processes mentioned in our dataset. • The automatic training procedure mentioned in this study causes the whole system to be adaptable to updating the dataset. • The present study is developed into software which enables the user to skip the training and testing process. [ABSTRACT FROM AUTHOR]