In this study, failure data of computer numerical control machine used in defense industry was analyzed to develope maintenance algorithm with a Markov feature. An imperfect preventive maintenance model that minimizes long-term operational cost is created for the machine wearing down randomly over time. The reliability-centric preventive maintenance policy was developed where the system status was monitored instantaneously. The use and age-related deterioration process of system is defined as the failure rate increase factor and age reduction factor, and these variables are combined to create hybrid failure model. As result of the imperfect maintenance algorithm developed for the multi-component machine, minimum long-term total unit cost, optimum system reliability value, number of maintenance and times between sequential maintenance cycles are obtained as outputs. Furthermore, system sub-equipment was specified that needs to be maintained in each cycle. Morover, imperfect maintenance activities are planned when the reliability level of subsystems drops to the predetermined R value.