Distribution system is part of power system which distribute electric power for local users and which is electrical system between the substations fed by the transmission system and customer meter. Distribution system reliability evaluation is a measure of continuity and quality of power supply to the consumers and also gives information about performance of the system and helps to know the cost or losses incurred by the utility and the customer due to power interruption. In this thesis in order to know the existing performance of Gonder radial distribution system historical reliability analysis is carried out using the collected data from North West region Ethiopian Electric Utility head office Bahirdar and Gonder distribution station for the years 2014/15 and 2015/16 G.G. The system reliability indices (SAIFI=346.574 int./yr, SAIDI=257.125 hr/yr , ASAI=0.985 %, EENS=629.97MWhr/yr and TCENS=409480.5 Birr/yr.) of the study area show that Gonder city distribution system is unreliable as compared to standard practices by other European countries and Ethiopia’s reliability requirement. Also which shows EEP’s distribution system is noncompliance with standards. And as a solution to mitigate this problem and improve the reliability of the power system supplementing the distribution system by distributed generations (DG) is investigated. Distributed generation is considered as an electrical source connected to the power system, in a point very close to/or at consumer´s site, which is small enough compared with the centralized power plants. Research work included by this thesis focuses on using an optimization methodology for identifying proper location for 0.3MW of DG units. To determine the optimal location and number of DG unit’s particle swarm optimization has been developed in terms of parameter selection, so as to obtain the minimum total cost of energy not supplied. The performance of this proposed methodology is tested on Gonder15KVdistribution feeder. Results show the efficiency and robustness of the proposed PSO algorithm to finding the very near optimal location of the DG units for enhancing the reliability of radial distribution system i.e. 32.57%, 21.03% and 32.51% improvement in SAIDI, TCENS and EENS respectively.