Power system consists of some power plants with variant types. Each power system has typical problem, but usually the main problem is inability power plant to perform as it is desired. Thus the power system could not serve the peak load reliably. There is a gap between installed capacity (design capacity) and real capacity called hidden capacity. The hidden capacity which operating in period hours called energy losses. This study would diagnose the root cause of energy losses using Bayesian Belief Network (BBN). Software Genie 2.0 is used to solve this problem. Based on data from power system planning dispatcher division, the most contribution energy losses came from coal fire power plant. The root cause of energy losses would be found quantitatively by considering its probability. This study also gives a quantitative prediction result that would be achieved and benefit that would be obtained when some scenarios were executed.