The accuracy of the Caltrac accelerometer was studied in 28 young (26.1 +/- 1.1 yrs) and 28 older (64.8 +/- 1.0 yrs; M +/- SEM) men and women to determine its usefulness as an activity monitor for young and older adults. Oxygen uptake was measured by indirect calorimetry while subjects walked on a motorized treadmill at six different speeds while wearing an accelerometer on the hip and on the upper back. The test-retest reliability of the Caltrac was r = .95 and r = .98 for young and older subjects, respectively. In the young group, activity counts correlated highly to net caloric expenditure (absolute minus estimated resting expenditure) (hip vs. kcal/kg: r = .89; back vs. kcal/kg: r = .88) when averaged across speeds, but the relationship was only moderate (back: r = .51; hip: r = .46) when the influence of speed was removed. In the older group the correlations were considerably weaker (back: r = .73; hip: r = .25), and essentially zero when the influence of speed was removed. These data indicate that the Caltrac is a highly reliable accelerometer that is useful for assessing qualitative differences in the level of physical activity among groups, but it lacks accuracy in quantifying energy expenditure in individuals, especially in older adults.