There are a number of benefits to making efficient load phase balance, such as loss minimization, energy restoration, security, reliability and voltage balance. Optimal load distribution is obtained by solving the load re-distribution problem as a combinatorial optimiz ation problem. This enables the best switching option that gives a balanced load arrangement among the phases and minimizes power loss to be arrived at. In this paper, we investigate the use of adding decaying self-feedback continuous Hopfield neural network (ADSCHNN) for load re-arrangement in the low voltage circuit of the distribution network. The network energy function of the ADSCHNN is constructed for objective function that defined the load balancing problem. The ADSCHNN is applied to solve the problem when load is represented in terms of current flow at the connection points. The results obtained using ADSCHNN are compared with those from a heuristic algorithm, and that CSA. Simulations results show that the ADSCHNN is very effective and outperforms other known algorithms in terms of the maximum difference of the phase currents.