Unreliable power distribution reduces user power consumption, affects daily activity and drags modern life style. Basically, Power Distribution Reliability has been a major challenge in Debre Markos town. Thus, the objective of the study is to assess the reliability of the current Debre Markos distribution system and to improve it. In this thesis residential, and commercial and industrial customer cost of electric power interruption in Debre Markos town is carried out. From the result the residential commercial and industrial customers faced to additional cost due to power interruptions. The historical reliability assessment of Debre Markos substation 15 KV distribution system for the year of 2006 and 2007 EC is carried out using analytical methods. From the two year report data the average number of power interruption frequency is 476, and the average interruption duration is 744 hours per year which indicates that the frequency of interruption is high and the duration of outage is long. The reliability of Debre Markos distribution system is improved by reducing SAIDI and EENS values using sectionalizer switches. The problem of optimal placement of sectionalizer switch is solved by using binary particle swarm (BPSO) optimization technique. The algorithm of binary particle swarm optimization is coded and simulated in Matlab software and illustrated in one of overhead 15 KV Debre Markos distribution feeder (FEEDER 4). The predictive reliability assessment is calculated by considering the existing sectionalize switches in the system and by placing optimal sectionalize switches in the feeder and enhance the reliability indices. Both manual and automatic sectionalizer switches are considered and automatic switches are preferable. The number and location of automatic sectionalize switches are identified to improve the reliability of system. From the test results, the proposed technique presents better performance, (27% reduction in EENS) in comparison with that of the existing system and SAID reduces from 285 to 216 hours/year (24% reduction) using optimal automatic sectionalizer switches over existing system.