A standard method for exploring item bias is the intergroup comparison of item difficulties. This paper describes a refinement and generalization of this technique. In contrast to prior approaches, the proposed method deletes outlying items from the formulation of a criterion for identifying items as deviant. It also extends the mathematical framework of item difficulty comparisons to allow the simultaneous analysis of any number of groups. As an example, the proposed method is applied to a set of quantitative items selected from a business school admission test. (Author)