The main drawbacks of mean-variance model are to generate corner solutions and low diversity in the portfolios. To overcome these defects, firstly, we propose a new proportion entropy function as an objective function to generate well-diversified portfolio. Secondly, considering the transaction cost and liquidity, we present a new fuzzy mean-variance-entropy multi-objective portfolio selection model to find tradeoffs between risk, return and the diversification degree of portfolio, which is able to address a more realistic portfolio selection problem. Thirdly, we combined several efficient schemes to form an efficient algorithm to maintain the diversity of obtained solutions and to solve the presented multi-objective portfolio selection model. The proposed multi-objective portfolio model combined with the multiobjective evolutionary algorithm can overcome these defects fundamentally. Finally, to demonstrate the efficiency and effectiveness of the proposed model and algorithm, the designed algorithm is compared with two famous algorithms: multi-objective evolutionary algorithm based on decomposition (MOEA/D) and non-dominated sorting genetic algorithm II (NSGA-II) through some simulations based on the data of the Shanghai Stock Exchange Market. Simulation results show that the proposed algorithm is able to obtain better diversity and more evenly distributed Pareto fronts than the other two algorithms, and our proposed portfolio model can yield good performance of portfolio.