Data literacy is complex. When students investigate the natural world, they must be able to gather data, organize it in tables and spreadsheets, analyze it in context, and describe and interpret it--usually as evidence to support a scientific argument. These skills are echoed in the science and engineering practices of the "Next Generation Science Standards": "Because raw data as such have little meaning, a major practice of scientists is to organize and interpret data through tabulating, graphing, or statistical analysis. Such analysis can bring out the meaning of data--and their relevance--so that they may be used as evidence" (NGSS Lead States 2013, Appendix F, p. 9). But before students can identify patterns in data or use it as evidence, they must be able to graph it. In 2007, The authors began working with scientists and teachers in Maine to explore students' data literacy skills. They found that when students began to organize, graph, and interpret their data, many were unsure about what kind of graph to make. Most made bar graphs, regardless of their research question. They also treated the graph like an end product in itself, instead of using it to see patterns and make arguments. Although students had the mechanical skills to generate graphs, they did not logically decide what kind of graph would best suit their particular research question. Consequently, the authors developed the Graph Choice Chart (GCC), a tool to help students choose the appropriate graph. This article describes the GCC and gives examples of how teachers have used it in their classrooms.