We proposed and tested a novel idiom priming paradigm in reading. While previous idiom priming studies measured response times in decision-based tasks or examined eye movements on idioms or their components, we recorded eye movements of 52 participants on 68 semantically related probes that followed either English idioms or unrelated control phrases (e.g., “You need to [bite the bullet]/[check the diary] and make this unpleasant decision immediately”). Idiom priming was observed on the late eye-movement measures on the probe (as predicted by the memory-based integration account, Colloway & Perfetti, 2017), validating the proposed paradigm for use in studies of idiom learning and processing. To test whether idiom priming is reduced across the clause boundary, in half of the sentences, probes were located in a different clause, separated with a semicolon. The priming effect was not reduced across the clause boundary, further supporting the memory-based integration account of idiom priming.