The main aim of the electric utilities is to provide a continuous and reliable supply without violating system constraints and operational limits. But during contingencies, system frequency and voltages get declined owing to real and reactive power deficiencies. These situations may lead to cascaded failures and complete blackout in the system. In order to reduce the risk of cascaded outage and blackout, load shedding has been considered as a preventive scheme. This paper presents a new music inspired harmony based optimization algorithm known as improved harmony search algorithm (IHSA) to find an optimal load shedding strategy for radial distribution systems during an overload contingency. The radial distribution systems are the final link of the interconnection between power systems and the consumers with unidirectional power flows. Overload contingency in the radial distribution systems without and with installed distributed generations (DGs) are the two cases considered in this paper. By the introduction of the distributed generations, the electrical distribution system has a locally looped system and bidirectional power flows. The main objective of the proposed algorithm is to minimize the sum of curtailed load based on their assigned degree of importance and system losses within the operational and security constraints of the system. The proposed method has been tested on IEEE 12-bus, 33-bus and 69-bus radial distribution systems. The feasibility of the proposed algorithm has been established and compared with genetic algorithm (GA) in terms of solution quality over realistic test systems considered. © 2015 Penerbit UTM Press. All rights reserved.