Annals of Operations Research 116, 225–242, 2002  2002 Kluwer Academic Publishers. Manufactured in The Netherlands. Review of Methods for Increasing Discrimination in Data Envelopment Analysis LIDIA ANGULO-MEZA and MARCOS PEREIRA ESTELLITA LINS Programa de Engenharia de Produção, COPPE/UFRJ,Caixa Postal 68507, CEP 21945-970, Rio de Janeiro, RJ, Brazil Abstract. We present a review of methods for increasing discrimination between efficient DMUs in Data Envelopment Analysis. These methods were classified into two groups: those that incorporate a priori information and those that do not use or minimize the use of such a priori information. We also compare these methodologies regarding their specific characteristics. Keywords: Data Envelopment Analysis, weight restrictions, Value Efficiency Analysis, cross-evaluation, super efficiency, multiple objectives model Introduction Since the appearance of Data Envelopment Analysis [5], this methodology has been applied in real life studies, revealing some drawbacks, amongst which we can highlight: (a) lack of discrimination among efficient DMUs that occurs when the number of DMUs is small in comparison with the total number of variables in the analysis; (b) unfitness of the weighting scheme, which frequently can be unreal, giving a big weight to variables with less importance or giving a small (or zero) weight to impor- tant variables; (c) multiple optimal solutions for the weighting scheme of extreme efficient DMUs. This review was made with emphasis on the first problem (a) bearing in mind that it is closely related to problems (b) and (c). Some of the methodologies reviewed are not designed specifically for the purpose of increasing discrimination, this being rather a side-effect. We classified the methodologies into two groups: the first group comprises those methods that incorporate a priori information provided by a decision-maker or expert into the model, while the second group of methods does not require such a priori infor- mation. Within the first group we considered three streams: weight restrictions, prefer- ence structure and Value Efficiency Analysis. Within the second group we present three methodologies: super efficiency, cross-evaluation, and a multiple objective linear pro- gramming approach.