Four experiments investigated whether mental imagery is involved in the comprehension of figures of speech. The general strategy was to employ similes that differed in the degree to which perceptual knowledge would be required for comprehension. "Perceptual" similes (A rope is like a snake) were expected to provide the strongest evidence for the use of perceptual information during comprehension. In contrast, so-called "mixed" similes (A lighthouse is like a candle) tested whether perceptual processing would be affected by the availability of other routes for comprehension. In Experiments 1 and 2, the similes were accompanied by pictures that were either consistent or inconsistent with the perceptual aspects of the similes' meanings. In Experiment 1, comprehension of the perceptual similes received the greatest facilitation from prior presentation of a consistent picture, but no difference was found in the degree to which the two picture types primed comprehension of the mixed similes. In Experiment 2, people were faster to verify the consistent pictures after hearing the perceptual similes, and this facilitation was present at ISIs of both 0ms and 750ms. For the mixed similes, some weak evidence for the activation of perceptual information was apparent early, but this effect disappeared after the delay. Experiment 3 ruled out the possibility that the results of Experiment 2 were due to the consistent pictures being more typical depictions of the target concepts. Finally, Experiment 4 used a visual interference technique to establish that the comprehension of perceptual similes is indeed accomplished through visual processing. For mixed similes, and for two additional groups of relatively abstract similes, no evidence consistent with visual processing was observed. Together, these experiments demonstrate that perceptual information can indeed play a role in figurative language comprehension, and that such information is apparently instantiated through visual mental images. These results motivate an account of "perceptual processing" that draws upon previous work in visual working memory and which has implications for language processing more generally. (LLBA 2000, vol. 34, n. 5)