Personal total exposure (E t ) is composed of exposure to pollutants of ambient origin (E a ) and nonambient origin (E na ), both of which are associated with health effects. However, E a is more relevant for estimating the health effects associated with ambient air pollutants. Associations between E a and ambient concentrations (C a ) were examined for different subpopulations in the Detroit Exposure and Aerosol Research Study (DEARS). First, E a was estimated for PM 2.5 , NO 2 , and O 3 . The associations between C a and E a were then characterized with mixed effect models. C a served as a good surrogate for E a for PM 2.5 in both summer (slope = 0.80) and winter (slope = 0.55), but only in summer for O 3 and NO 2 (slope = 0.13 for NO 2 , and slope = 0.03 for O 3 ).The effects of exposure determinants on E a –C a associations were examined using mixed effect models. Exposure determinants examined include personal activities, building and household characteristics and ambient apparent temperature (AT). The strength of associations between C a and E a differed between that for the entire study population and different subpopulations, which in turn differed from each other. Personal activities and building characteristics significantly affected the E a –C a associations through modifying the air exchange rate (AER). C a was not a significant indicator of E a for O 3 and NO 2 for homes with central AC or with windows closed. Ambient AT affected the E a –C a associations through modifying both AER and personal activities. The fraction of daily time spent outdoors peaked (6%) at about 22 °C, and monotonically decreased when ambient AT departed from 22 °C. Adequate accounting for the effects of exposure determinants on the E a –C a associations will lead to a better understanding of the E a –C a associations and of the uncertainties associated with using ambient concentrations as surrogates for personal exposures of ambient origin.