A number of hypotheses were tested concerning elements of Graduate Record Examinations (GRE) items that might affect the performance of blacks and whites differently. These elements were characteristics common to several items that otherwise measured different concepts. Seven general hypotheses were tested in the form of sixteen specific hypotheses, most with five or six items. Items were developed in pairs, some with and others without the hypothesized element, and were administered within the GRE analytical, verbal, and quantitative sections. Log linear analyses of the sixteen hypotheses showed interactions between group membership and item version, indicating a differential effect on black and white examinees' performance. A latent trait approach was also used to examine individual item bias for eight of the hypotheses. The manipulations included context clues, item format, vocabulary, selecting the false rather than the true answer in a multiple choice item, understanding the item writer's inference, test wiseness, location of the correct response, and ambiguity versus structure. It was concluded that the item manipulations did have differential effects on blacks' and whites' performance. However, the complexity of the effects suggested that other uncontrolled factors affecting performance were present. (Author/GDC)