Combining the characteristics of disassembly line, a mathematical model was established for multi-objective DLBP where several optimization objectives were considered, such as minimizing the number of workstations, balancing idle time at each workstation, removing hazardous, high demand components as early as possible and minimizing the disassembly direction changes,then, an improved ant colony optimization algorithm was developed. In this algorithm, a hybrid search mechanism by comprehensively considering utilization, exploration and random search was constructed. The remove time, the harm and the demand of parts of DLBP were collaboratively considered and used for the heuristic information of the proposed algorithm. Finally, the computer experiments on different scale test instances were performed and the results indicate the feasibility and validity of the proposed algorithm.