Abstract: In order to describe the uncertainty of the returns on a given portfolio, the portfolio selection problems were investigated by using possibilitic theory. Firstly, the expected returns and risk were given by using probabilistic mean and variances, respectively. A portfolio optimization problem whose object is maximizing the return on an investment and minimizing risk was built. Secondly, the non-linearity and multi-objective of the model was considered and solved by genetic algorithm which incorporated the elite and mutation operations. Finally, the proposed scheme was verified by an empirical analysis of the model with a data set from Shanghai and Shenzhen stock market, and the results show that the method is effective.