Studies investigating the effects of aging on nonliteral language processing have mainly focused on one sensory modality, for example written vignettes. In the current study, we used a video-based task to examine the effect of healthy aging on social communication perception using a novel database called RISC (Relation Inference in Social Communication). By means of an online recruitment platform, we asked young, middle-aged, and older adults between the ages of 18 and 76 (N = 100) to evaluate videos of actors using different forms of literal and nonliteral language, such as sarcasm or teasing. The participants' task was to infer the speakers' belief and the speakers' intention. Older participants demonstrated lower accuracy in discriminating nonliteral from literal interactions compared to younger and middle-aged groups. When evaluating speaker intentions, older adults judged sarcasm as friendlier compared to literal negative utterances. We also found that the older the participant, the more difficulty they have identifying teasing as insincere. Our results expand on age-related similarities and differences in evaluating speaker intentions and demonstrate the practicality of the RISC database for studying nonliteral language across the lifespan.