This research examined how perceived honesty affects face memory. Photographs of college men and women were rated for honesty, and then selected exemplars of honest and dishonest faces were shown singly for study as targets. In the test phase, four types of four-face arrays were shown: (a) honest targets with honest distracters, (b) dishonest targets with honest distracters, (c) honest targets with dishonest distracters, and (c) dishonest targets with dishonest distracters. In addition, half of the four-face arrays involved four new faces (target absent). The primary effect of honesty in the target-present arrays was on confidence judgments, with subjects being more confident of their decisions about dishonest targets. In the target-absent arrays, there were more false alarms when honest-looking alternatives were present, contrary to what would be expected if a criminal stereotype was operating. A second experiment using only target-present arrays found better recognition and greater confidence associated with dishonest faces. An examination of additional ratings on this set of faces for typicality, attractiveness, and likability, revealed that overlapping features may be more responsible for these effects than the operation of a criminal stereotype.