Due to the harsh fact that existing bilateral matching required preference ordinals of opposite side members are provided by each agent, a novel method is developed. Agents provide evaluation in form of the intuitionistic fuzzy set through a pairwise comparison of alternatives, then generating an intuitionistic fuzzy preference relation of agents when applied to whole. By use of the improved minimum logarithmic square algorithm, the satisfaction degree of each agent is calculated based on the intuitionistic fuzzy preference relations(IFPR). Furthermore, a single-object optimization model with stability constraint is developed, which maximizes satisfaction degree on both two sides’ agents. The results can be obtained by solving this model. Finally, a numerical example is given to illustrate the feasibility and effectiveness of this method.