Several questionnaires are available for assessing physical activity, but few of these instruments have been validated, particularly with respect to energy balance. Twenty-one healthy men 28-55 years old participating in a free-living, controlled feeding experiment completed eight widely used physical activity questionnaires. These were compared with measured caloric intake and resting energy expenditure during a period of stable body weight. Physical activity indices or daily energy expenditure estimates derived from the questionnaires generally increased with energy intake. The questionnaires were moderately well correlated with each other; interquestionnaire correlation coefficients ranged from 0.09 to 0.81 (median = 0.53). Correlations between the questionnaires and energy intake, which ranged from 0.13 for the Minnesota Leisure Time Activity instrument to 0.49 for the Harvard Alumni questionnaire, were higher than between the questionnaires and nonresting energy expenditure; that is, energy intake minus resting energy expenditure (correlation coefficient range 0.05-0.32). The Five-City Project questionnaire yielded an average estimate of total caloric expenditure that most closely approximated intake (96%). These data indicate that although estimates of individual energy expenditure or physical activity may be less than optimal, most of the questionnaires evaluated provide reasonable group means for these parameters.