PM(Preventive Maintenance) can avail the generating unit to reduce cost and gain more profit in a competitive supply-side power market. So, it is necessary to perform reliability analysis on the power systems in which reliability is essential. Thus, to schedule optimal PM planning based on reliability that is defined RCM(Reliability-Centered Maintenance), FMECA(Failure Mode Effects and Criticality Analysis) assessment is very important. Therefore, in this paper, the procedure of FMECA assessment for optimal RCM is proposed by probabilistic approach using real historical failure data of combustion-turbine generators in Korean power systems. The stochastic FMECA is performed based on the effects of probable failure modes of combustion-turbine generating unit.