In many real problems, it does not consider the ranking characteristic of criteria to make multi-criteria decision with classical rough sets theory. By Dominance-based Rough Sets Approach (DRSA), this paper develops methods for weight assignment respectively under two settings, with decision sorting classes and without decision sorting classes. Compared to present methods, the advantage of our methods is that: from descriptive set of the given problem, extracts the inherent rules implied by alternatives dominance relations with respect to each attribute, and avoids excessively depending on experts' experience; realizes completely data driving. Finally an example illustrates our methods are effective and reasonable.