Evaluation is an inherent part of education for an increasingly diverse student population. Confidence in one's test-taking skills, and the associated testing environment, needs to be examined from a perspective that combines the concept of Bandurian self-efficacy with the concept of stereotype threat reactions in a diverse student sample. Factors underlying testing reactions and performance on a cognitive ability test in four different testing conditions (high or low stereotype threat and high or low test face validity) were examined in this exploratory study. The stereotype threat manipulation seemed to lower African-American and Hispanic participants' test scores. However, the hypothesis that there would be an interaction with face validity was only partially supported. Participants' highest scores resulted from low stereotype threat and high face validity, as predicted. However, the lowest scores were not in the high stereotype threat/ low face validity condition as expected. Instead, most groups tended to score lower when the test was perceived to be more face valid. Stereotype threat manipulation affected Whites as well as non-Whites, although differently. Specifically, high stereotype threat increased Whites' cognitive ability test scores in the low face validity condition, but decreased them in the high face validity condition. Implications for testing and classroom environment design are discussed. (Contains 6 tables and 1 figure.)