Four experiments demonstrate that priming effects depend on the context-appropriate meaning of the prime words. Most studies of semantic construct activation have presented prime words in contexts where the meaning of each word was invariant (e.g., word puzzles). in this research, we used words in contexts that supported either literal or figurative meanings, and found that only the context-appropriate meanings had subsequent priming effects on person-perception judgments. In Experiment 1, participants read the word "fire" in one of three contexts: a figurative use that implied recklessness ("playing with fire"), a figurative use referring to a hot streak ("on fire"), or a literal use ("playing by the fire"). Differential priming effects were obtained in a subsequent person-perception task that were consistent with the context-appropriate meanings of the priming expression. In Experiment 2, a conventional idiom, "break a leg," produced divergent priming effects when used idiomatically than when used literally. In addition, we found evidence for inhibition of irrelevant literal meanings in Experiments 3a and 3b that provided support for the role of inhibitory processes in metaphor and idiom comprehension. Implications for how figurative language might differentially activate knowledge structures and the role of inhibitory processes in social perception are discussed.