Abstract: This paper proposes a rough set model of logical difference operation of grade and precision, and defines the concepts of rough set regions. In the new model, basic structure of rough set regions is obtained by transformation formulas between variable precision approximations and graded approximations. Regular algorithm and structural algorithm are proposed and analyzed to calculate rough set regions, and application prospect of the new model in decision table is explored and obtained. The new model has extended graded rough set model and classical rough set model.