When it comes to graphing data, most professionals show little method or creativity. They typically limit themselves to a small repertoire of graph types and select from it on the basis of habit, if not sheer ease of production. Similarly, the many books on graphing devote much attention to graphical integrity and readability, but little or none to graph selection. We developed a methodology to help engineers, scientists, and managers choose the "right graph" on the basis of three criteria: the structure of the data set in terms of number and type of variables, the intended use of the graph, and the research question or intended message. The first and third criteria allow one to construct an effective two-entry selection table.