This study sought to improve students' comprehension of scientific graphs by adapting scaffolding techniques used to aid text comprehension. In 3 experiments involving 121 female and 88 male college students, some students were shown cognitive aids prior to viewing 4 geography graphs whereas others were not; all students were then asked to write a summary of the main information in the graphs. Students who received signaling or structural graphic organizers, which are designed to facilitate the cognitive process of organizing, generated more relational statements (e.g., "the deeper the water, the greater the sediment concentration"; d = 0.60 and d = 1.10, respectively) but not more causal statements (e.g., "because the river is picking up sediment from the banks") than did students in control groups; students who received concrete graphic organizers, which are designed to facilitate integration with prior knowledge, generated more causal statements (d = 0.39) but not more relational statements. Combining signaling and concrete graphic organizers resulted in increases in both relational and causal statements (d = 0.76 and 0.93, respectively). Results are consistent with a cognitive model of graph comprehension.