In the age of modern industries, reliabilitiesand suitable performances of precidingly designed systems are very important factors because of systems complexity and considerable cost of systems failure or systems redesign.Using the advantage of impelementing reliability optimization strategies such as increasing the reliability of components or redundancy allocation to subsystems, the systems could performe more reliable. The usual approach in reliability theory is to consider all of the components as binary-state elements. While, in the real wolrd, many systems performe their task with degraded performance levels. This phenomenon is mainly caused by the degredation of components and parts in the system or/and the failure of system elements which deteriorates the system performance. This type of system is called multi-state system (MSS). In this thesis, the reliability optimization problem for fuzzy multi-state systems (FMSS) is solved using redundancy allocation strategy. The main difference of this research in comparison to previoulsly done researches is the consideration of uncertainty for performance levels and their probabilities in a multi-state system. Moreover, after the recognition of the optimized structure of a system, the availability diagram together with its upperbound and lower bound is ploted in order to uncertainty interval. Genetic algorithm (GA) and differential evolution algorithm (DE) are used to solve the presented optimization models. The results showed that the DE algorithm is more powerful method in comparision with GA in order to solve redundancy allocation problem.