Biometric systems, which use people?s physiological characteristics for identification or authentication, have become increasingly popular for countering fraud and other misdeeds. Identification is about finding the best match between a person and a group of possible candidates (for example, finding a suspect in a criminal investigation). Authentication determines if people are who they claim to be. Identification results in a list of possible matches (with an indication of the degree of match); authentication returns a binary decision. In this article, the authors describe a face recognition system that uses a new distance measure for authentication. The system performs matching on a fusion of multiple views of each person. Although the authors focus on optimizing the performance of single-modal operation, their proposal is generic enough for combining other modes of operation (for example, speech analysis). The system?s results have been very encouraging.