This study explores three formats for the presentation of tangible line graphs to visually handicapped people, using the capsule-paper production process. The formats are L/no-grid (standard x and y axes/data curves on smooth background), box/no-grid (duplicated axes enclose graph space on all four sides/smooth background), and box/grid (duplicated axes/data curves superimposed on grid). Subjects answered questions concerning the coordinate value of points, the difference in coordinate value between two points, and configurational aspects of the data. Questions that involved determining coordinate values were answered most accurately from the box/grid format, followed by the box/no-grid, and, lastly, the L/no-grid. Response times did not differ. A different pattern emerged with questions concerning configurational aspects of the graph. With these there was no difference in accuracy between the formats, but the L/no-grid format led to faster responding than the box/no-grid, which in turn was faster than the box/grid. The practical implications of these data are considered.