Abstarct: Restructuring and recent developments in the power system and problems arising from construction and maintenance of large power plants have raised amount of interest in distributed power generation (DG) source. Distributed generation units due to specifications, technology and location network connectivity can improve system and load point reliability indices. This demonstrates the importance of finding the optimal size and placement of DGs. Because of the radial structure of distribution networks, in the event of an error occurring due to the impossibility of recovery the load, long-term interruption are applied to customers. Distributed generation units can decrease interruption duration and number of consumers that are influenced by errors. In this thesis, the allocation and sizing of Distributed Generators (DG) in distribution networks are determined using optimization. Placement is done with numbers and varying sizes of distributed generation resources.The objective function of the proposed method is to improve the reliability indices such as system average interruption duration index (SAIDI), system average interruption frequency index (SAIFI) and average energy not supplied per customer index (AENS) at lowest cost. The optimization is baxsed on a Shuffled Frog Leaping Algorithm (SFLA) aiming to determine optimal DG allocation and sizing in distribution network. The SFLA is a new memetic mexta-heuristic algorithm with efficient mathematical function and global search capability. To evaluate the proposed algorithm, the 34-bus IEEE test system, is used. In addition, comparative studies of the genetic algorithm have also been performed. The results show that this proposed method is one of the best methods to improve service reliability, reduce the customer outage costs and decrease the power cost.