How do we distribute attention to interactive biological motion (BM)? There are 2 main hypotheses: (a) distribution-by-individual hypothesis, suggesting that interactive BM will not be taken as one unit in attention distribution, and an individual BM is independently selected; and (b) distribution-by-group hypothesis, indicating that interactions between BM can integrate them as one attention unit. We examined these hypotheses using a modified cueing paradigm. Participants observed 4 upright BM interacting in pairs (paired condition) or not interacting (unpaired condition), and after a cue for 1 agent, they determined whether the probe was the letters T or L (Experiment 1, sample size = 20). The results demonstrated better performance for probes presented on BM in the same interaction compared to BM equidistant but in different interactions. These differences were not present in the unpaired condition or for inverted BM (Experiment 2, sample size = 20), excluding a possible influence of low-level features. The findings indicate that interactive BM can serve as the elementary unit of attention in accordance with the distribution-by-group hypothesis. (PsycInfo Database Record (c) 2020 APA, all rights reserved).