This study used a novel research paradigm to examine gait control during real-time between-person collision avoidance. Ten young adults (M = 20.1 ± 1.52 years) were required to walk across a six metre simulated pedestrian crossing, while avoiding a collision with one or two oncoming pedestrians. The potential for social interaction was manipulated by having the oncoming pedestrians walk with (2MP) or without (2P) looking at a mobile phone. Participants took longer to complete the crossing when avoiding a collision with two oncoming pedestrians (2MP: M = 5.68s; 2P: M = 5.74s) in comparison with baseline (M = 4.96s). Gait velocity decreased and was more variable when avoiding a collision during the 2 P condition, whilst the anterior-posterior separation distance between pedestrians and the participants at the initiation of peak mediolateral deviation was significantly smaller in 2MP compared to 2P. These findings offer preliminary understanding on how gait control may be adapted to changes in the availability of other persons’ gaze orientation information. Future work is needed to further understand how different adaptive behaviours emerge relative to other persons during pedestrian crossings.