Distribution Systems (DS) are usually structured as weakly-meshed but the majority of them operate with a radial topology, mainly in order to accommodate the protection coordination. Obtaining the optimal radial configuration under several criteria has been an active research topic for more than two decades. Because of the computational burden and the non-linearity of the problem, the majority of the proposed methods and techniques, single or multi-objective, use various meta-heuristics. The DS reconfiguration problem, respecting the radiality constraints, is formulated in this paper as a multi- objective Mixed-Integer Linear Programming (MILP) problem. An adequate representation of the Pareto set is produced using an improved implementation of the e-constrained method. The objective is to determine the optimal radial configuration during several time intervals, minimizing the active power losses and the cost emerging from the switching operations. The proposed methodology is tested using a 16-node sample system.