Background: Standardized methods to measure allergen exposure are essential to assess the relationship between exposure, sensitization, and asthma. Most studies have measured allergen levels in reservoir dust, although air samples may be more representative as a measure of inholed allergen. The aim of this study was to define the relationship between mite, cat, and dog allergen content in the reservoir dust and the levels in the ambient air. Methods: Dust samples from the living-room floor (LF) and sofa (S) were collected in 127 homes: 62 without and 65 with pets (31 dogs, 34 cats). Air samples were taken in the same room, with a high-volume pump for 1 h (flow 60 l/min). Der p 1, Fel d 1, and Can f 1 were determined by mAb- based ELISA. Results: Airborne Der p 1 was below the detection limit (0.8 ng/m3) in all homes, with reservoir levels (GM and range) being 1.14 μg/g (0.2-66) and 1.15 μg/g (0.2-127) in LF and S, respectively. Airborne Can f 1 was detected in 40/62 homes without pets (range 0.6-12.4 ng/m3) and in all homes with dogs (range 0.5-99 ng/m3). In the multiple linear regression analysis, Can f 1 level in the LF was an independent correlate of the airborne Can f 1 (P=0.01, homes with dogs; P=0.04, homes without dogs). Fel d 1 was detected in the air in 16/62 homes without pets (range 0.16-1.8 ng/m3) and in all homes with cats (range 0.4-22.3 ng/m3). Fel d 1 level in the LF was an independent correlate of the airborne Fel d 1 in homes without cats (P=0.008), but airborne levels in homes with cats did not correlate with reservoir levels. Conclusions: The aerodynamics of each allergen must be taken into account when assessing exposure: while levels in reservoir dust are the best available index for mite allergens, airborne levels might be more suitable for defining exposure to pets. If air samples are difficult to obtain, levels of Can f 1 and Fel d 1 in the LF samples should be used as a surrogate measure of personal exposure.