The optimal configuration of location and capacity of the distributed generation (DG) can ensure its better technical and economic utility. Based on the detailed analysis on peculiarity of DG, this paper establishes a model of DG which takes the minimal active network loss of DG and the least cost into consideration. A bacterial colony optimization (BCO) algorithm based on the basic growth law of bacterial colony is presented. An information sharing method is built for the colony. This BCO algorithm provides a new type of termination; it will terminate the iteration when the colony vanishes, regardless of the iteration number or the precision value. Finally, the effectiveness of the proposed algorithm is illustrated by experiments.