This study evaluated the effect of behavioral modeling and social factors promoting stair use. Alternating baseline and intervention phase experimental design. San Diego International Airport, San Diego, California. Stair use was coded for 15,574 filmed participants. This study compared the effects of three types of behavioral modeling: natural models (i.e., passersby), single experimental model (i.e., confederate), and confederate model pairs providing verbal prompts. Variables were coded based on systematic observation of videotapes, including demographics, day and time, and the following indicators of physical and social reinforcement contingencies: dress, luggage, children, social group, and speed. Reliability ranged from .64 to .88. Bivariate and logistic regression models stratified by gender. Stair use increased over baseline by 102.6% with no model present and by 61.8% in the presence of natural models for men and women (p < .001). Controlling for multiple covariates, the odds ratios for stair use ranged from 1.76 to 2.93 for men and from 1.82 to 2.54 for women across the levels with natural and confederate models present (all p < .001). Modeling can prompt stair use, and findings for social and environmental reinforcement contingencies support the Behavioral Ecological Model. Modeling may explain partial maintenance of stair use in public areas after removal of prompts (e.g., signs, banners). Results inform interventions for increasing physical activity as part of daily routines.