<p>Both task-relevant and -irrelevant information in working memory (WM) automatically biases attentional selection toward sensory inputs. While semantically congruent audiovisual presentation has been shown to enhance interactions between memory-relevant features and ongoing visual perception, it remains unclear whether and how such bimodal presentation modulates the guidance of attention by memory-irrelevant features. To address this issue, we employed a WM/visual search dual-task paradigm, in which participants performed a gap-directed search task involving a memory-irrelevant feature while retaining the other feature for later retrieval. The memory-irrelevant feature was presented either via bimodal (audiovisual) or unimodal (visual, Experiment 1; auditory and visual, Experiment 2) during the WM encoding phase. The results of Experiment 1 showed that during the middle stage of visual search, the audiovisual condition produced larger capture effects and response time (RT) benefits than the visual condition, with an earlier onset of these RT effects. This can be attributed to the multisensory facilitation rather than a mere advantage of auditory memorandum (Experiment 2). We propose that multisensory interactions in WM enhance attentional resources, which are integrated with task demands to facilitate both the magnitude and temporal dynamics of attentional biasing by memory-irrelevant features. These findings provide new insights into the influence of congruent audiovisual stimulation on attentional guidance by irrelevant WM contents, thereby elucidating multisensory interactions within the memory-driven attention framework.</p>