Two studies investigated the effect of stereotypes held by a prospective audience on participants' reactions to a stereotype-disconfirming member. In Study 1, participants formed an impression of a positive disconfirming gay in order to communicate it to an audience known to hold a negative versus positive stereotype about gays. As predicted, participants subtyped the deviant more in the former than in the latter case. Moreover, participants' stereotype at the end of the study mirrored the audience's assumed stereotypes about gays. In Study 2, participants learned about a stereotype allegedly held by an ingroup or an outgroup audience about Belgians and then received information about a Belgian who disconfirmed the stereotype. As predicted, the deviant was seen as less typical when he violated the stereotype held by an ingroup than by an outgroup audience. Also, participants' stereotype about Belgians was more similar to the one held by the ingroup audience. A mediational analysis confirmed that participants subtyped the disconfirming member in order to embrace the stereotype advocated by the ingroup audience. Results are discussed in light of recent models of stereotype change. Copyright © 2006 John Wiley & Sons, Ltd.