As the problem of three-phase unbalance in a distribution network with distributed generations (DGs) is prominent, a multiobjective dynamic reconfiguration method for three-phase balance is proposed to make reconfiguration more reasonable and effective. Based on three-phase power flow calculation and reconfiguration process analysis of distribution networks with DGs, a new reconfiguration model with comprehensive optimization objectives of minimizing three-phase unbalance factor and number of switching times are established. A network connectivity discrimination method based on the algebraic connectivity of graph theory is adopted to quickly remove infeasible solutions, and a new multiobjective molecular differential evolution algorithm is designed to improve optimization depth and avoid prematurity in the optimization process, which is based on the principle of closer molecules with greater intermolecular repulsion. The multiobjective optimal dynamic reconfiguration of a modified IEEE 34-bus test feeder with DGs is carried out and the simulation results demonstrate the effectiveness and superiority of the proposed method.