The validity of the Caltrac movement sensor for use with preschool children was assessed. Caltrac-derived values for energy expenditure were compared with those derived via laborious coding of direct observation that involved classification of the child’s videotaped activity every other 5 seconds for an hour in the day-care center or on the playground. Both Caltrac and direct observation values were expressed in kilocalories. The subjects were 20 children with a mean age of 35 months. The correlation coefficient for the total of indoor and outdoor activity was r= .62 (p<.01). The separate correlations for indoor and outdoor activity were r=.56 (p<.05) and r=.48 (p<.05), respectively. However, when the children’s weight, height, age, and sex were factored out of both the Caltrac and direct observation scores, the correlations fell to r= .25 (n.s.), r= .47 (p<.05), and r=.16 (n.s.) for the total, indoor, and outdoor activity, respectively. Thus the Caltrac seemed to record indoor activity (mainly walking) more accurately than it recorded the more varied playground movements, casting doubt on its value as a means of measuring physical activity in children 2-3 years of age.