The interconnection of renewable energy sources with distribution systems is attracting increasing interest because these renewable sources are inexhaustible and nonpolluting. Wind and photovoltaic are among the most mature of these energy sources, and their penetration continues to increase. In this paper a method based on GA (genetic algorithm) is presented to investigate the distribution system reconfiguration problem taking into consideration the effect of load variation and the stochastic power generation of renewable DG (distributed generators units). The presented method determines the annual distribution network reconfiguration scheme considering switching operation costs in order to minimize annual energy losses by determining the optimal configuration for each season of the year. The uncertainties related to DG power and varying load are considered by the creation of a probabilistic generation-load model that combines all possible operating conditions of the renewable DG units with the probability of their occurrence, followed by the incorporation of this model into the reconfiguration problem. The constraints include the voltage limits, the line current limits, the radial topology, and feeding of all loads. In order to evaluate the effectiveness of the proposed method, both balanced and unbalanced distribution systems are used as case studies. (C) 2013 Elsevier Ltd. All rights reserved.