ERF proteins are essential transcriptional regulators of different cellular functions such as response to biotic and abiotic stresses, metabolism and development. They belong to the AP2/ERF gene superfamily of transcription factors and are characterized by the presence of a conserved AP2/ERF DNA binding domain in their sequence. A detailed phylogeny based classification of the ERF genes is important because it gives insights into the function of the different members and sub-groups of the family. We check here the classification of the Arabidopsis thaliana ERF protein family based on the NJ tree built from the protein distance matrix proposed by Nakano and collaborators, employing Bayesian Inference (BI) with the substitution model fitted to the coding sequences of the conserved DNA-binding AP2/ERF domain. We expanded the dataset used by Nakano et al. for phylogeny reconstruction by including the so called VI-like and X-like groups. Our analysis could not confirm the close affinity of these groups to the groups VI and X, suggested in the original analysis so we propose to treat them as new groups XI and XII. Groups VI and IX could also not be confirmed, as they appeared as paraphyletic assemblages in our analysis. All other groups suggested by Nakano et al. appeared as monophyla also in our BI tree. Our findings will contribute to the functional characterization of the genes belonging to the ERF family.