Phased-mission system (PMS) is the system subject to multiple, consecutive and non-overlapping tasks. Much more complicated problems will be confronted when the PMS is repairable since the repairable system could perform the multi-phases mission with more diversity requirements. Besides, various maintenance strategies will directly influence the reliability analysis procedure. Most researches investigate those repairable PMSs that carry out the multi-phases mission with deterministic phase durations, and the mission fails once the system switches from up to down. In this case, one common maintenance strategy is that failed components are repairable as long as the system keeps in up state. However, many practical systems (e.g., construction machinery, agricultural machinery) may be involved in such multi-phases mission, which has uncertain phase durations but limited by a maximum mission time, within which failed components can be unconditional repaired, and the system can be restored from down state. Comparing with the former type of repairable PMS, the latter will also concern phase durations dependence, and both the system and components included have the state bidirectional transition. This paper makes new contributions to the reliability assessment of repairable PMSs by proposing a novel SEFT-MC method. Two types of repairable PMS mentioned above are considered. In our method, a specific sequence-enforcing fault tree (SEFT) is proposed to correctly depict failure logical relationships between the system and components included. In order to transfer the graphical fault tree (no matter its size and complexity) into a modular reliability model used in Monte Carlo (MC) simulation, an improved linear algebra representation (I-LAR) approach is introduced. Finally, a numerical example including two cases corresponding to the two types of repairable PMS is presented to validate the proposed method.