This study investigated the validity of the Physical Activity Interview (PAI) for assessing children's self-reported physical activity. Third and fifth graders wore an accelerometer and were interviewed using the PAI to determine 12-hour recall accuracy for activity expressed as energy expenditure (EE). Caltrac estimates of EE for the 12-hour day (12-HEE) and activity EE were the validation criteria. 12-HEE correlations were significant (p<.001) between Caltrac and PAI for third (r=.79) and fifth grade (r=.80). Caltrac and PAI Total Activity were significant for fifth grade (r=.82, p<.001) but not for third grade (r=.36, p<.021). Analysis of tertiles based on the Caltrac demonstrated that the PAI significantly (p<.002) discriminated between activity levels for fifth graders but not for third. The use of PAI may be age-dependent.