In　this　paper,　we　propose　a　novel　artificial　bee　colony(ABC)　algorithm　for　the　multidimensional　knapsack　problems,　which　introduces　the　attraction　pheromone　and　presents　a　transition　strategy　based　on　the　attraction　pheromone.　In　our　algorithm,　the　scout　generates　a　food　source　according　to　the　transition　strategy　and　replaces　the　abandoned　food　source　by　comparison　with　the　corresponding　elite　food　source,　while　the　employed　bee　and　onlooker　modify　the　repair　operator　using　the　transition　strategy　in　the　determination　of　the　neighborhood　of　a　food　source.　Experimental　results　show　that　our　approach　performs　better　in　the　quality　of　solutions,　the　convergence　speed　and　the　time　performance　than　traditional　ABC　algorithm.