The present research deals with the opportunities for the availability predictive modeling of a thermal plant using the Markov process and probabilistic approach. These opportunities will be identified by evaluation of a power generation system of a thermal power plant. This feasibility study covers two areas: development of a predictive model and evaluation of performance with the help of the developed model. The present system under study consists of four subsystems with three feasible states: full working, reduced capacity working and failed. Failure and repair rates of all subsystems are assumed to be constant. After drawing a transition diagram, differential equations are generated and then a probabilistic predictive model using Markov approach has been developed, considering some assumptions. The availability matrix for each subsystem is also developed, which provides various availability levels for different combinations of failure and repair rates of all subsystems. On the basis of this study, the performance of a power generation system is evaluated. The developed model helps in the comparative evaluation of alternative maintenance strategies.