This study examines the relative effectiveness of four different social network representations for improving human problem-solving accuracy and speed: node-link diagrams, adjacency matrices, tables, and text. Results suggest that visual network representations improve problem-solving accuracy and speed, compared with text. Among the visual representations, tables produced superior problem-solving outcomes for symbolic tasks and link-node diagrams produced superior problem-solving outcomes for spatial tasks. These results partially support a cognitive fit model of problem-solving support. There is not “one best way” to represent network data. Instead, it is important to match network representations and problem-solving tasks.