Occupants can significantly influence both the heating energy requirements and the indoor air quality of a building by opening and closing doors and windows. If the effects of these actions are to be accurately estimated, both the quantity and character of these exchange flows must be determined. In this paper, data on gravity-driven exchange rates through open doors obtained from field experiments at the Alberta Home Heating Research Facility are compared with laboratory model simulations and theoretical predictions. Experimental results are presented which show that simple theoretical models provide both physical insight and an accurate estimate of air flow through doorways. Tracer gas techniques and thermocouple arrays were utilized in full scale experiments to determine air exchange rates and gravity current frontal velocities. These tests were conducted over a range of indoor-outdoor temperature differences from 3°C to 45°C. Experiments were also performed using a 1:20 scale model with salt water mixtures to simulate indoor-outdoor temperature differences. In spite of a Reynolds number mismatch of a factor of 7, the model and full scale exchange flow rates and gravity current frontal velocities were in very good agreement when opening times were adjusted using a buoyancy time scale based on Froude number similarity.