In many-objective optimization, maintaining a good balance between convergence and diversity have turned out to be a considerable challenge for classical evolutionary algorithms. The large scale solution set required to describe the entire Pareto optimal front hinder the decision makers from finding the most satisfactory solution, whereas he/she is only interested in a limited part of the objective space. The dilemma can be handled by incorporating the preference information in the search process. In this paper, an evolutionary algorithm based on region preference is proposed for many-objective optimization. The preference model is constructed in combination of the target region and reference points. To focus the search on the preference region while maintaining well convergence and diversity within the region, a tri-level ranking criterion is introduced into the proposed algorithm, and different rank works at different phase of the search process. A fuzzy theory-based interactive approach is proposed to guide more individuals to further search into the objective space with higher preference degree and help the decision maker choose the most interested solution. The proposed algorithm has been extensively compared with other state-of-the-art preference-based algorithms on DTLZ1~ DTLZ4 test problems having 3-10 objectives. The experimental results indicate that the proposed algorithm can achieve competitive and better performance. Moreover, we extend the algorithm to handle multiple target regions.