While most of the work on resource scheduling in Grid computing has focused on optimal makespan of the whole systems, our work addresses this problem from the resource consumer's point of view by providing multi QoS constraints and related methods. An algorithm of resource evaluation and selection, RESA, is presented, applying multiple attribute decision making (MADM) and hierarchical analysis methods. It has a hierarchical and scalable QoS model. And the optimal result is more objective because of combining prediction and preference information. © 2010 IEEE.