With the development of new materials and advanced structural analysis, alongside increasing aesthetic requirements, recent years have witnessed a trend toward longer, taller, and lighter footbridges. Different from vehicular bridges, footbridges carry relatively small service loads and are more susceptible to vibrations, due to their lower stiffness, damping, and modal mass. More often than not, vibration serviceability limit state governs the design of footbridges. Providing an accurate evaluation of vibration serviceability performance of existing bridges requires techniques that can include modeling and measurement uncertainties. In this paper, a population-based method called error-domain model falsification (EDMF) is used to assess the vibration serviceability for two pedestrian bridges: Fort Siloso Skywalk located in Singapore and the Dowling Hall Footbridge located at Tufts University in the United States. The unknown properties of the footbridges are identified using the ambient vibration data measured on site. This method is also compared with two other data-interpretation methodologies, that is, residual minimization and traditional Bayesian model updating. The findings show that, through explicitly accounting for measurement and modeling uncertainties, EDMF can provide more accurate identification and prediction results for vibration serviceability assessment of pedestrian bridges.