Objectives. We investigated whether individuals mimic the stair/escalator choices of preceding pedestrians. Our methodology sought to separate cases where the 'model' and 'follower' were acquaintances or strangers. Design. Natural experiment. Methods. Infrared monitors provided a second-by-second log of when pedestrians ascended adjacent stairs/escalators in a mall. Manual timings established that stair climbers spent ≥ 7 s on ascent, during which time they could act as models to following pedestrians. Thus, individuals who mounted the stairs/escalator ≤ 7 s after the previous stair climber were assigned to a 'stair model' condition. A 'no stair model' condition comprised individuals with a gap to the previous stair climber of ≥ 60 s. The stair model condition was subdivided, depending if the gap between model and follower was 1-2 s or 3-7 s. It was hypothesized that the former cohort may know the model. Results. Percentage stair climbing was significantly higher in the 'stair model' versus 'no stair model' condition (odds ratio [OR]= 2.08). Subgroup analyses showed greater effects in the '1-2 s' cohort (OR = 3.33) than the '3-7 s' cohort (OR = 1.39). Conclusions. Individuals appear to mimic the stair/escalator choices of fellow pedestrians, with more modest effects between strangers. People exposed to message prompts at stair/escalator sites are known to take the stairs unprompted in subsequent situations. Our results suggest that these individuals could recruit a second generation of stair climbers via mimicry. Additionally, some of the immediate behavioural effects observed in interventions may be a product of mimicry, rather than a direct effect of the messages themselves.