Extracting slopes from arrays of visual features is crucial for interpreting graphs. To understand broader influences on slope perception, we investigated the effects of concurrent sounds, of relative graph location, and of semantic priming on a visual search task in which observers searched for a graph with a positive or negative slope. Four bar graphs or scatter plots were simultaneously presented in separate quadrants of a visual display. Participants pressed a key as quickly as possible if one of the graphs displayed the target slope and otherwise refrained from response. A concurrently presented ascending pitch slowed responses to negative-slope targets, and concurrently presented descending pitch slowed responses to positive-slope targets, indicating crossmodal interference. This interference was eliminated when the sounds were presented 200 ms before the graphs, consistent with crossmodal interaction rather than response bias. Positive slopes were detected slowest in the upper-left quadrant whereas negative slopes were detected slowest in the upper-right quadrant, suggesting that slope detection is impeded when a graph is placed inconsistently with a mental number-line representation (negative values on the left and positive values on the right). Finally, positive slopes were detected faster when the search display was immediately preceded by a briefly flashed word "uphill" compared to the word" downhill" (and the converse for negative slopes), indicating a semantic priming effect, but this was observed only for scatter plots (the stimulus specificity precluding response bias). In summary, perception of visual slope is systematically influenced by auditory signals, by location of graphs, and by semantic priming.