A novel distribution network reconfiguration algorithm, named core schema genetic shortest-path algorithm (CSGSA) is proposed in this paper. It is based on the loads combination method. CSGSA consists of two steps: (1) searching for the optimal power supply paths for a sequence of loads one by one using shortest-path algorithm, and forming a core schema chromosome using core schema algorithm (CSA); and (2) searching possible core schema chromosomes for the optimal one using global optimization method-genetic algorithm. CSGSA is used for solving the loss minimum reconfiguration problem, and its global searching capability is greatly improved by using a proposed branch weight formulation, which considers the effect of residual branch capacities; CSGSA is also employed to solve the supply restoration problem and promising results have been obtained. The proposed method can be used as an efficient tool for distribution network loss minimization and supply restoration.