The load demand in Taiwan power system has been increasing during the past decade due to rapid economic growth on the island. To meet the ever-increasing demand, effective expansion plans for distribution systems are essential. The increasing load density can be met by either increasing the capacities of existing substations or planning a new substation. The load allocation problem arising from the latter approach will be addressed in the paper. The purpose of a load allocation plan is to reallocate some loads in the study system to the new substation such that certain desirable features of the distribution system such as minimum loss, minimum investment cost, high reliability, satisfactory voltage regulation, etc., can be met. Several approaches based on mathematical programming have been reported in the literature for the load allocation problem. However, these approaches require considerable computational effort and can not take many practical constraints into account. After an intensive interview with system planners in Taiwan Power Company, it was found that current practice of load allocation is a combination of hand calculations on such items as losses, voltage regulation, investment cost, reliability, etc. and some heuristic rules, which has been established based on past experience. The use of these rules in the utility motivated the development of a rule-based expert system for the load allocation problem. Considerable progress in the application of artificial intelligence (Al) technology to power system problems has been achieved during the past decade.