Faces with a stereotypic criminal appearance are remembered better and identiﬁed more often than other faces according to past research. In the present project, a random sample of police lineups was evaluated using the mock witness paradigm to determine whether criminal appearance was associated with lineup choices. In Study 1, mock witnesses were either provided with a description of the culprit or they were not. Participants also self-reported why they had selected a given face. In Study 2, the line-up faces were rated with respect to criminal appearance, distinctiveness, typicality and physical similarity. Criminal appearance was the primary reason self-reported for face selection in the no description condition. Mock witness choices in the no description condition were associated with only criminal appearance. When provided with a description, mock witnesses based their choice on the description. These ﬁndings are discussed in relation to lineup fairness.