The performance of the Caltrac accelerometer was studied in elementary school-age children under field and laboratory conditions. In Study 1, 35 children (20 boys, 15 girls, mean age = 10.8 yr) wore the accelerometer and a heart rate (HR) monitor for 2 d. Caltrac activity counts per hour were compared to the mean "activity HR", which was calculated by subtracting the mean of the five lowest HRs of the day from each recorded HR. Pearson r's between accelerometer and activity HR were 0.54 (P less than 0.001) on day 1 and 0.42 (P less than 0.02) on day 2. Inter-instrument reliability in the field was r = 0.96. Both accelerometer and HR data were significantly correlated with physical activity recalls of the same day. In Study 2, 15 children walked/ran for 10 min at 3, 4, and 5 mph on a horizontal treadmill while wearing two accelerometers. Oxygen uptake was directly measured each minute. Reliability of the Caltracs in the laboratory was 0.89. Activity count correlated r = 0.82 (SEE = 23%) with net calorie cost per kg of body weight. Net caloric expenditure per kg of weight was 0.101 kcal.kg-1.d-1 per Caltrac activity count. These data support the use of the Caltrac accelerometer as a physical activity measure for school-age children, and the objective data tended to corroborate the children's short-term activity recalls.