Of 67 office workers 27 (40%) had documented tuberculin skin test conversions after an estimated 4-wk exposure to a coworker with cavitary tuberculosis. Worker complaints for more than 2 yr before the tuberculosis exposure prompted investigations of air quality in the building before and after the tuberculosis exposure. Carbon dioxide concentrations in many parts of the building were found to be above recommended levels, indicating suboptimal ventilation with outdoor air. We applied a mathematical model of airborne transmission to the data to assess the role of building ventilation and other transmission factors. We estimated that ventilation with outside air averaged about 15 feet3/min (cfm) per occupant, the low end of acceptable ventilation, corresponding to CO2 levels of about 1,000 ppm. The model predicted that at 25 cfm per person 18 workers would have been infected (a 33% reduction) and at 35 cfm, a level considered optimal for comfort, that 13 workers would have been infected (an additional 19% reduction). Further increases in outdoor air ventilation would be impractical and would have resulted in progressively smaller increments in protection. According to the model, the index case added approximately 13 infectious doses (quanta) per hour (qph) to the office air during the exposure period, 10 times the average infectiousness reported in a large series of tuberculosis cases. Further modeling predicted that as infectiousness rises, ventilation would offer progressively less protection. We conclude that outdoor air ventilation that is inadequate for comfort may contribute to airborne infection but that the protection afforded to building occupants by ventilation above comfort levels may be inherently limited, especially when the level of exposure to infection is high.