In the first experiment, subjects watched a videotape of a simulated purse snatching involving an assailant and an accomplice. After a 1-week delay, each subject was presented two photoarrays in succession. One photoarray was of the assailant and the other was of the accomplice. Following each identification task, each subject who made an identification was asked to describe actions of the person he or she selected as having been involved in the crime. In the second experiment, the incidence of UT was studied using a live staged event involving an altercation between a student and a professor. Little conclusive evidence for UT was obtained using either the videotaped event or the live staged event. Negative cross-race effects, however, were observed, as well as null relations between both witness-reported confidence and witness-reported goodness-of-view with accuracy. High "self-monitors" showed the greatest probability of hits from target-present photoarrays, whereas politically conservative subjects showed the greatest probability of false alarms from target-absent photoarrays. Implications of the findings for the evaluation of eyewitness evidence are discussed. 42 references and 3 tables