The performance of passive chilled beams is characterized in this study to understand their physical behavior when applied in real indoor environments. Two full-scale experimental studies were conducted to map the performance of passive chilled beams in two different experimental settings: (i) single passive chilled beam testing in a controlled laboratory environment and (ii) multiple passive chilled beams testing in a real open plan office setting. The experimental results were then used to develop regression models for predicting the total cooling capacity and chilled surface temperature of single and multiple passive chilled beams, as a function of water flow rate, water supply temperature, air temperature above the chilled beam and area-weighted uncooled surface temperature in the space. The developed models showed good agreement with experimental results and can be used in building energy modeling tools for system simulation using passive chilled beams. Finally, it was found that the conventional method of predicting the total cooling capacity of a passive chilled beam from individual beam laboratory tests or manufacturers' catalogs may significantly underestimate the system performance in multi-unit configurations. These differences could influence optimal system sizing and commissioning and should be considered in future studies. (C) 2017 Elsevier B.V. All rights reserved.