This paper presents a novel approach for the reconfiguration of large-scale power distribution systems using genetic algorithms to improve the system's efficiency while minimizing power loss. The methodology was coded in DIgSILENT power factory, since this is widely used for power systems analysis in electric utilities and by researchers. Through an interconnection-sectionalizing switch incidence matrix, the methodology searches in the feasible topological solution space; also, operating conditions such as power balance, voltage level, line loading, and the type of switch actuated (remote-controlled or manual) are considered. The effectiveness of the method is demonstrated through simulation on a real distribution system of a Colombian utility. The results show that the proposed method can significantly improve the system's performance and reduce power loss. The proposed approach is a practical tool for helping network operators achieve efficiency in their power distribution systems.