Abstract
Sixty-two witnesses observed a target person and two days later either verbally described or constructed a facial composite of the target using the Mac-A-Mug Pro facial composite system. The composite production was preceded by either a guided memory or a standard police interview. The resulting composites were of low quality (not at all similar to a photograph of the target) and were of no value in selecting the target from a group of similar-looking people. It was suggested that the Mac-A-Mug Pro system is not useful for realistic witness settings, in which witnesses must construct the composites from memory. Suggestions are provided for modifying facial composite systems to make them more compatible with the psychological processes mediating face recall.