Although ever-growing literature has documented that a multisensory environment influences individuals' learning, there is less agreement as to whether it facilitates or hinders and why. In this work, we addressed this issue with a systematic review and meta-analysis and examined the multisensory facilitation and interference of learning. The present study reviewed the literature on audiovisual learning and quantitatively synthesized a total of 29 studies with 198 comparisons to (1) assess the effect of multisensory context by examining the role of congruency, (2) examine study-level factors-modality, age, and familiarity-that may moderate the effects observed across studies, and (3) discuss possible explanations and implications for theories of attention in learning. The results indicated that the multisensory effect was constrained by the congruency between multimodal information. Compared to a unisensory presentation, a multisensory context enhanced the task performance in the congruent conditions whereas it hindered the performance in the incongruent conditions. Furthermore, the effect of congruency was modulated by the modality of the target stimulus and the age of participants, and the multisensory effect was limited for familiar multimodal associations under incongruent conditions. Importantly, interference was found from the auditory distractors to the visual targets but not from the visual distractors to the auditory targets, in child participants but not in adults, and for the familiar multisensory associations but not for the unfamiliar ones. These results point to a critical role of selective attention in multisensory learning and imply roles of attentional focusing and filtering in processing congruent and incongruent multimodal information.