California Environmental Protection Agency. In order to evaluate more precisely the potential health risks from air pollution, it is critically important to have accurate estimates of the volume of air breathed (VE) by exposed populations. There is a substantial amount of characteristic VE data available in the literature, but values typically are representative of adult males during rest and in light to moderate activities. This is due in part to difficulties in the direct measurement of VE in free-ranging people, which requires portable respirometers that can restrict normal performance of some common activities. Thus, VE for free-ranging activity has usually been estimated from unobtrusive heart rate (HR) measurements in individuals whose VE to HR relationship response to varied intensities of cycle ergometer or treadmill exercise had been established in the laboratory. However, it is not known how accurately VE is predicted in the field when using this method. To resolve this problem, both VE and HR must be measured simultaneously in the field across a wide range of free-ranging activities. The primary purposes of this research were to 1) identify mean values and ranges of VE for specific activities and populations, and 2) develop equations which would permit VE predictions based on known activity and population characteristics. The subject population utilized in this study comprised 160 normally active individuals of both genders, and of varied age (6-77 years) and ethnicity. In addition, 40 children (6-12 years) were recruited for data validation and 12 young children (3-5 years) were identified as subjects for pilot testing purposes. Subjects completed resting (lying, sitting and standing) and active (walking and running) laboratory protocols, and usually one or more field protocols (i.e., play, car driving/riding, car maintenance, yardwork, housework, mowing and/or woodworking). Collected laboratory data included steady-state measurements of VE, HR, breathing frequency (fB) and oxygen consumption (VO2), while data collection in the field was limited to the continuous measurement of VE, HR and fB during each protocol. Resting responses for the children's groups revealed no significant gender differences and those for the adult groups demonstrated minimal age-group differences; therefore, resting data were combined into children, adult female and adult male groups. Heart rate and VE responses were poorly correlated in all resting postures for each group. However, fB was a better predictor for VE, with body surface area (BSA) being an important additive variable in multiple regression equations. Very similar observations were obtained from the cross-validation children's group. Regression analysis revealed higher r values for walking and running protocols than for field protocols, which were higher than for resting protocols. Typically, HR was poorly correlated with VE except during active laboratory protocols, whereas fB and BSA were variables that better predicted VE across all types of activities and population groups. The inclusion of all three variables (BSA, HR and fB) in multiple regression equations, generally provided the most accurate predictions of Vs across groups and activity types. The lower precision of prediction for active field protocols than for laboratory walking and running protocols was at~ibuxed primarily to the wide variety of individual activities and intensity of effort during field protocols. Using the mean VE values obtained for each population group and activity, field protocols were categorized into one of the following: sedentary activity, and light or moderate exercise. For the children's groups, spontaneous play protocols were identified as moderate exercise. Car driving/riding was classified as sedentary activity for both genders. Car maintenance for the male adult group was categorized as light exercise, while their VE responses to woodworking, mowing and yardwork protocols were class