The modality-match effect in recognition refers to superior memory for words presented in the same modality at study and test. Prior research on this effect is ambiguous and inconsistent. The present study demonstrates that the modality-match effect is found when modality is rendered salient at either encoding or retrieval. Specifically, in Experiment 1, visual and auditory study trials were either randomly intermixed or presented in blocks, followed by a standard (old-new) recognition test. The modality-match effect was observed for the mixed but not the blocked condition. Experiment 2 used a modality-judgment test (requiring a seen, heard, or new judgment). The resulting measure of recognition memory exhibited the modality-match effect for both list conditions. These results imply (a) that the modality-match effect is a consistent finding when modality is salient and (b) that the effect arises at retrieval rather than encoding. (Contains 4 tables, 1 figure and 2 footnotes.)