Reaching “win–win” settlement is the desired outcome of mediation. Logrolling is a strategy for achieving “win–win” trade-off. In this study, a logrolling strategy in mediation is proposed through which parties can improve the joint value by bargaining exchange and get convergence along the efficient frontier. A multi-objective decision making (MODM) model is employed to propose the efficient frontier and assist parties to engender “win–win” settlement. To operationalize the logrolling strategy, a web-based logrolling system is developed to assist parties to achieve “win–win” settlement in a user-friendly environment. The system includes 3 processes: reality test, preference identification and logrolling. Reality test is proposed to test parties' concession rate. Preference identification assists parties to identify their utility value of the bargaining alternatives. Logrolling is to provide user-friendly strategies for parties to make efficient trade-off that involves (1) when to concede (2) on which issue (3) for which party and (4) how much should be conceded. Finally a mock mediation experiment was conducted to examine whether the logrolling system can assist parties to achieve “win–win” settlement, where the system simulates a Mediator in action. The results are evaluated by comparing the difference between the mediator's expected logrolling outcomes and the subjects' actual logrolling outcomes. The logrolling-difference degree (L-DD) is used to measure this difference. It is found that the average of L-DD in bargaining range and reaching agreement are 11.43% and 8.46% respectively, which indicates that the logrolling system is having good potential in assisting parties to achieve “win–win” settlement.