MG workflow can be defined as the composition of manufacturing activities which execute on heterogeneous and distributed manufacturing resources in a well-defined order to accomplish a specific goal. The execute process of MG workflow includes model building, service finding, service selecting, service binding and service invocation. Service selecting is one of the most critical components in MG workflow. So, this research brings forth AHP-based grid nodes choosing method, which takes time, quality, cost, service, fidelity, reliability, policy (TQCSFRP) as rules, and chooses best resource node as target. The dynamic service binding and invocation between an activity request and an optimal service provider is realized when a workflow is in execution. Higher flexibility and self-adaptability of workflow management are achieved. Furthermore, the application architecture is depicted at the end of the paper to illustrate the utilization of MG workflow.