The Implicit Relational Assessment Procedure (IRAP) was designed to examine implicit beliefs or attitudes. In Experiment 1, response latencies obtained from Irish participants on the IRAP showed a strong preference for Irish over Scottish and American over African. In contrast, responses to explicit Likert measures diverged from the IRAP performance in indicating Irish equally likeable to Scottish and African more likeable than American. Using a similar IRAP, Experiment 2 showed that participants from the United States showed strong implicit preferences for American over Irish, Irish over Scottish, and Scottish over African; the explicit Likert measures again diverged from the IRAP. The findings provide preliminary support for the IRAP as a useful measure of implicit beliefs.