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.