On the basis of representing the packing components and space with a new type of octree data structure,this paper presents an effective heuristic method for arbitrary shaped 3D-packing problems. By forming the octrees of packing space and packing objects in the same octree family, the octree of each object can be looked as a node of octree of packing space in certain level,and a series of heuristic rules are used to find a"best fitnode" for each object octree in space octree.Finally,a hypothesis example with 8 arbitrary shaped objects is given and the calculation process is described in detail,and the obtained result is very promising.