Design of a fuzzy logic power system controller with satisfactory performance is not an easy task. The difficulties come from two aspects. First, design of a fuzzy logic computer mainly uses the experience of the human experts. To acquire enough heuristic knowledge from the domain experts and to represent this kind of knowledge appropriately with a set of fuzzy rules presents difficulties. Second, it is difficult to appropriately tune the parameters used in the fuzzy logic controller. These parameters are commonly determined by a "trial and error" method which is rather time consuming. In this paper, a genetic algorithm is introduced to design an optimal fuzzy logic controller. The proposed method has been used to design an optimal fuzzy logic excitation controller for a generating unit. Test results with the fuzzy logic controller show very satisfactory results.