This experiment examined correspondence bias associated with the expression of pro- or anti-American sentiment in the months following September 11, 2001. Participants read a pro- or anti-United States essay written by a person whose name was varied to suggest that he either was or was not a Muslim. Participants were informed either that the position taken in the essay had been chosen or assigned. Inferences made about the essay writer's true opinions demonstrated strong and consistent correspondence bias only when participants believed the essay writer was not a Muslim. When participants believed the essay writer was a Muslim, correspondence bias was diminished on one measure and disappeared completely on another. These findings are consistent with the notion that participants were concerned about rushing to incorrect conclusions about a Muslim target person. The strong correspondence bias exhibited by participants making judgments about a non-Muslim is consistent with Gilbert and Malone's (1995) assertion that unrealistic expectations lead to correspondence bias.