Repairable systems with latent failures are systems when a failure is identified only at special inspection times, appearing with a given fixed period. In the case of its occurrence the restoration process is launched at the earliest follow-up inspection time. Maintenance policies for such systems can be categorized into two main categories: calendar based and age based policies. In the calendar based policy, the system testing and maintenance is realized at a fixed date, whereas in the age based policy we suppose that maintenance actions are realized at a fixed time interval from the latest time of maintenance. In the age based policy, a real calendar time of inspection or preventive maintenance is random in fact, due to random occurrences of failures and random duration of restorations. Relevant issue of the paper is to recognize differences between both kinds of maintenance policies. The paper summarizes innovative stochastic reliability models, recently developed and used for unavailability quantification of repairable and maintained systems with latent failures. Models are transformed into numerical codes to answer the relevant matter: to what extent is affected unavailability in context with age-based versus calendar-based maintenance policy? The differences of unavailability, computed in the course of time, between age-based and calendar-based maintenance policies are demonstrated on examples selected from both references and practice.