Abstract: The problem of hierarchical multiobjective optimization of industrial large scale systems is studied. It is classified into three basic forms and three new approaches are presented. The original problem is simplified to a coordinator and a series of multiobjective subproblems with lower dimension by decomposition and coordination principle and multiobjective optimal technique, so that only less computer memories are needed and the convergence is speeded. The approaches are also simple, practical and efficient. These have beeh shown by several numerical examples.