The subjects were 30 college undergraduates who took part in pairs in what was described as a study of personality and risk taking. They were asked to play a card game involving money. Without their knowledge, they had been randomly assigned to the roles of victim or bystander prior to starting the game. During the competition, a male confederate flung open the door, looked around, rushed toward one of the subjects, asked whether the money belonged to the subject, and grabbed the money and ran without waiting for an answer. The entire event lasted about 3.5 seconds. The subjects were then interviewed about the event, asked to view a seven-photo display, and asked to identify the culprit if he was shown in the display. Although the bystanders as a group were slightly more accurate than the victims in their descriptions, a significant gender difference emerged. The female bystanders were significantly more accurate than the female victims, whereas the male bystanders were slightly, but nonsignificantly, less accurate than the male victims. Although 25 of the 30 witnesses made an identification from the photograph, only 8 made accurate identification. These eight witnesses were all bystanders. Self-reports of nervousness showed that it was not an explanatory factor. The greater arousal of victims or the bystanders' greater tendency to focus on the criminal's face might account for the difference. Possible practical implications and the need for further research to establish the generalizability of the observed victimization effect are discussed. One table, footnotes, and 28 references are supplied. (Author abstract modified)