This work presents a novel multi-objective bee swarm optimization (MOBSO) method. The proposed method divides a swarm as experienced foragers, onlookers and scouts. An adaptive windowing mechanism is used by the experienced foragers in order to select their own leaders and adjust their next positions. Also, the adaptive windowing is used for truncating the most crowded members of the archive. A new way is proposed in which the scouts and adaptive windowing are used to maintain diversity over the Pareto front. A scout creates a hypercube using knowledge provided by a pair of archive members, and flies spontaneously in it. The provided knowledge by the experienced foragers is used by the onlookers in order to adjust their flying trajectories. The proposed algorithm was compared with existing multi-objective optimization methods. The experimental results indicate that the proposed approach not only presents a uniformly distributed Pareto front but also identifies results with greater accuracy.