Abstarct: Nowadays, organizations are facing with various projects and investment opportunities. Regarding to the existing constraints and activity of the business environment, the survival and success of organizations to achieve competitive advantage requires opting for an appropriate and limited group of projects. This selection faces two fundamental challenges. The first one is concerned with the conditions of decision making in the problem of project portfolio selection. Since most decision makers find themselves under the conditions of uncertainty and ambiguity and even with incomplete information to make a decision, considering the exact amounts or the distribution function for the parameters of a project may be impractical and non-functional.1 The second challenge is concerned with the tools and methods of optimization that have been used up to now. Among existing approaches, the evolutionary algorithms such as GA and PSO are known as the most fundamental tools of optimization. Using these tools, there is a fundamental problem and that is in implementing each prior mexta-heuristic method, determining special control parameters are required. Choosing the controlling parameter amounts in a right way is very sensitive and also influences on the algorithm performance and the final answer as a result. In addition, the improper selection of parameters increases the calculation process or may result in a local answer. This study considers the uncertainty conditions in the project portfolio selection model by using the fuzzy concepts and then tries to enter the self-optimization from the engineering literature to the management literature for the first time with a modern approach in order to pass the second challenge and implement it in project portfolio selection. The considered technique does the optimization operations without the necessity to determine the special controlling parameters and it determines the general optimization. This technique has been used only in Electrical and Mechanical Engineering for optimizing the designing variables.