Based on the analytical hierarchy process proposed a virtual machine migration evaluation model for green cloud computing. Through the establishment of index system of virtual machine CPU task usage, the construction of pair-wise comparison matrix and the adoption of five-point rating scale, the model is able to evaluate the liability of migration and resource usage of the virtual machine. By using relevant attributes of virtual machine, an attribute clustering based collaborative filtering method is applied for the calculation of similarities between virtual machines and the generation of migrate recommendation. Finally it can help the cloud server administrators for choosing the optimum green virtual machine migration scheme to save power consumption of cloud server cluster. Experiments show that, the proposed model and method is feasible and has the vital significance to the green cloud computing.