This study examined how 12- and 13-year-old students' mathematics and science background knowledge affected line graph interpretations and how interpretations were affected by graph question levels. A purposive sample of 14 students engaged in think aloud interviews while completing an excerpted Test of Graphing in Science. Data were collected and coded using a rubric of previously cited factors, categorized by Bertin's ("Semiology of graphics: Diagrams, networks, maps." The University of Wisconsin Press, Ltd., Madison, 1983) theory of graph interpretation. Data analysis revealed responses varied by graph question level. Across levels, students interpreted graphs in one or more of the three ways: mathematical word problems (focusing on an algorithm), science data to be analyzed (incorporating science knowledge), or no strategy. Although consistently used across levels, the frequency and usefulness of approaches varied by question level.