An ignored stimulus is later recognized at enhanced levels if it had previously been aligned with a target from a separate task. This has been demonstrated using both visual and auditory presentations. Here we extend these findings to multisensory conditions. Participants were required to detect immediate repetitions in a sound or picture stream while ignoring superimposed words presented in the opposite modality (either written or spoken, respectively), and then underwent a surprise recognition test for these words. Contrary to the previous unisensory examples (Dewald, Sinnett, & Doumas, in press; Dewald & Sinnett, 2012), a significant difference between recognition rates for target-aligned and non-aligned words was not observed. However, a highly significant difference in response latency was observed, with target-aligned words being responded to much more quickly. This finding was robust and observed when the surprise test was presented in either the visual or auditory modalities, as well as across modalities.