We used existing data to develop distributions of time-averaged air exchange rates (AER), whole-building 'effective' emission rates of volatile organic compounds (VOC), and other variables for use in Monte Carlo analyses of U.S. offices. With these, we explored whether long-term VOC emission rates were related to the AER over the sector, as has been observed in the short term for some VOCs in single buildings. We fit and compared two statistical models to the data. In the independent emissions model (IEM), emissions were unaffected by other variables, while in the dependent emissions model (DEM), emissions responded to the AER via coupling through a conceptual boundary layer between the air and a lumped emission source. For 20 of 46 VOCs, the DEM was preferable to the IEM and emission rates, though variable, were higher in buildings with higher AERs. Most oxygenated VOCs and some alkanes were well fit by the DEM, while nearly all aromatics and halocarbons were independent. Trends by vapor pressure suggested multiple mechanisms could be involved. The factors of temperature, relative humidity, and building age were almost never associated with effective emission rates. Our findings suggest that effective emissions in real commercial buildings will be difficult to predict from deterministic experiments or models.