On the current research of bi-level programming, a new approach to get its fuzzy satisfactory solution with floating code genetic algorithm is proposed, The approach provides the decision makers (DMs) with a group of inexact optimal solutions, so that in the process of interavtive comparison, evalua-tion and selection, a satisfactory solution can be got easily. Moreover, the approach can give the DMs more information about the decision environment and can adapt to the changes of the DM's preference, which makes the decision process more reasonable and flexible．