This study evaluates the test-retest reliability and validity of self-report measures of physical activity that can be self-administered in classroom settings to 4th grade students. Four different self-report formats were tested on 66 students. To assess test-retest reliabilities, self-report measures were administered on two occasions, separated by a 3-day interval between Time 1 (Friday) and Time 2 (Monday). One-way model intraclass reliabilities ranged from .51 to .74. Three days of monitoring with the Caltrac accelerometer were used as the validity criterion. Only one of the three weekly recalls, the Weekly Activity Checklist, was supported by significant validity correlations at both Time 1 (r = .34, p < .01) and Time 2 (r = .26, p < .05). The 1-day recall, Yesterday Activity Checklist, correlated significantly (r = .33, p < .01) with the previous day's Caltrac monitor score. Although two of the physical activity recall formats were found to be superior to two others, these data highlight the limitations of children's self-reports. Two self-report formats were found to have modest levels of reliability and validity with 4th grade children when administered in a classroom setting.