In this paper, we address a distribution network reconfiguration problem (DNRP) that operates on standard configurations of electrical networks. The main objectives handled by the DNRP are the minimization of power loss, the minimization of the number of switching operations and the minimization of the deviations of bus voltages from their rated values. Due to its multiobjective nature and combinatorial aspects, the DNRP is considered as NP-hard. Hence approximate approaches are very promising in generating high quality solutions within a concurrential run time. In this paper, we develop a distribution network reconfiguration approach using an iterated local search (ILS) algorithm, known to be a powerful stochastic local search method. This has been investigated and illustrated on an IEEE 33-bus and IEEE 69-bus radial distribution systems. Indeed, we proposed a novel solution encoding that avoids in a smooth and natural way the creation of isolated components and closed loops, in each generated network configuration.