A new preference-based method is presented to determine optimal schedule of maintenance actions for deteriorating bridges. The maintenance planning of deteriorating bridges is formulated as a multi-objective optimisation problem involving the maintenance cost as well as the condition grades of the bridge deck, girder and pier. The maintenance cost corresponds to life-cycle maintenance cost that is defined as the sum of the total expenses caused by the maintenance interventions during the life span of the bridge. The condition grades of the bridge components are characterised by time-dependent performance profiles that have been constructed by statistical regression with the actual condition grade data of bridges in Korea. To effectively address the multi-objective optimisation problem, a genetic algorithm is adopted as a numerical searching technique of multiple optimal solutions. As for the explored optimal solutions, a preference-based decision making process is then introduced to select reasonable maintenance scenario that most fits the designer's preference while balancing the mutually conflicting design objectives. The effectiveness of the proposed method is verified numerically on a typical pre-stressed concrete girder bridge. The numerical results demonstrate that the optimal maintenance scenario determined by the proposed method can be a well-balanced maintenance strategy in terms of cost and bridge performances.