Abstarct: Nowdays, the portfolio optimization problem is a necessary and important issue for participation of the investors in the economic and financial development of each country. By idolizing the Markowitz model, investors are looking for profit and higher return and at the same time minimizing the risk. In this thesis, at first, we introduce the proposed Markowitz model and the other portfolio selection models that are baxsed on two criteria named, return and risk. Subsequently, we investigate multiobjective portfolio selection problem baxsed on additional measures, such as liquidity and annual dividend, ... apart from return and risk. Moreover, due to insufficient information available about assets and inaccuracy of return, risk and liquidity values, we consider the portfolio selection problem with interval and fuzzy coefficients and express different solution methods for optimizing each of these issues. Then, we explain the fuzzy multiobjective portfolio selection problem baxsed on credibility measure and present a new method to solve it. In this way, we combine fuzzy goal programming with a hybrid intelligent algorithm, that includes fuzzy simulation and real-coded genetic algorithm. This method is very useful for when the fuzzy parameters are in any general functional form expect linear, triangular or trapezoidal, because, it will be difficult the conversion of fuzzy problem into its crisp equivalent. However, we present a new approach baxsed on mextaheuristic methods to consider fuzzy parameters of the portfolio selection problem when they are in the triangular or trapezoidal form. To this end, we use the Harmony search algorithm, that moves to the optimal spaces baxsed on possibility approach. By solving the portfolio problem via this algorithm, we achieve to a set of Pareto optimal solutions.