Many types of visual representations are used in math textbooks but not all of them contain mathematically relevant information. Little research directly addresses the effects of different types of representations on mathematics performance. Theories offer differing perspectives about how visual representations such as illustrations influence student learning. Here, we investigated the effects of diagrams and contextual illustrations on trigonometry problem solving. Diagrams helped all students, but the effect of contextual illustrations depended on students' backgrounds. Additionally, not all subgroups of students accurately assessed the effect of illustrations on their performance. We emphasize the need to consider how different types of visual representations interact with student characteristics and the problem-solving task. [For the complete proceedings, see ED584829.]