Zimbardo's deindividuation hypothesis was reexamined by individuating some subjects. Twelve four-person groups administered shock to a confederate in a 2 × 2 × 2 factorial design with two levels of social individuation (subjects' level of anonymity to group members), two levels of nonsocial individuation (subjects' opportunity to give identifying information to the experimenter on a word association test), and two levels of individual differences. It was found that subjects who individuated themselves by giving information to the experimenter on the word association test showed more, not less, antisocial behavior. Furthermore, this increased antisocial behavior was exhibited only by externally oriented subjects: those with an external locus of control and a low mysticism score. Some of those more antisocial subjects were also more likely to report feeling like they stood out from the group. It was concluded that antisocial behavior in this paradigm may thus be a response to experimenter demand felt most strongly when subjects felt most identifiable.