Purpose: Physical activity questionnaires (PAQs) are considered the most co
st-efficient method to estimate total energy expenditure (TEE) in epidemiol
ogical studies. However, relatively few PAQs have been validated using doub
ly labeled water (DLW) in women or in samples with diverse ethnic backgroun
ds. This study was conducted to validate the Arizona Activity Frequency Que
stionnaire (AAFQ) for estimation of TEE and physical activity energy expend
iture (PAEE) over 1 month using DLW as a reference method. Methods: Thirty-
five relatively sedentary women completed the AAFQ before participating in
an 8-d DLW protocol to measure TEE. TEE and PAEE were estimated from the AA
FQ by calculating resting metabolic rate (RMR) using the equation of Miffli
n et al. (AAFQ(mif)), by measuring RMR using indirect calorimetry (AAFQ(ic)
), and using MET conversion (AAFQ(met)). A predictive equation for TEE was
generated. Results: The mean SD for TEE and PAEE from DLW were 9847 +/- 255
5 kJ.d(-1) and 5578 +/- 2084 kJ.d(-1), respectively. Formulas using RMR to
calculate the TEE and PAEE from the AAFQ tended to underestimate TEE and PA
EE, whereas those that included only weight tended to overestimate TEE and
PAEE. On the basis of the Mifflin et al. equation, the AAFQ tends to undere
stimate PAEE by 13%. This underestimation may be. explained by the low lean
body mass of the sample population and by effectiveness of the METs/RMR ra
tio in the obese. The following predictive equation was calculated: TEE (kJ
.d(-1)) = (86.0 * average total daily MFTs) + (2.23 * RMRmif)-6726. When th
e predictive equation is used, TEE calculated from the AAFQ is highly corre
lated with DLW TEE (adjusted r(2) = 0.70, P<0.001). Conclusion: The AAFQ is
an effective tool for the prediction of TEE and PAEE in epidemiological st
udies.