This experiment used functional Magnetic Resonance Imaging to examine the relation between individual differences in cognitive skill and the amount of cortical activation engendered by two strategies (linguistic vs. visual-spatial) in a sentence-picture verification task. The verbal strategy produced more activation in language-related cortical regions (e.g., Broca's area), whereas the visual-spatial strategy produced more activation in regions that have been implicated in visual-spatial reasoning (e.g., parietal cortex). These relations were also modulated by individual differences in cognitive skill: Individuals with better verbal skills (as measured by the reading span test) had less activation in Broca's area when they used the verbal strategy. Similarly, individuals with better visual-spatial skills (as measured by the Vandenberg, 1971, mental rotation test) had less activation in the left parietal cortex when they used the visual-spatial strategy. These results indicate that language and visual-spatial processing are supported by partially separable networks of cortical regions and suggests one basis for strategy selection: the minimization of cognitive workload.