There is need to develop low cost, practical, and accurate measures of physical activity in children and adolescents, and self-report is a promising methodology for children that is applicable for large studies. The purpose of the present study was to assess the reliability and validity of several self-reports of physical activity. Subjects were 36 fifth-, 36 eighth-, and 30 eleventh-grade male and female students. The test-retest reliabilities were r = 0.77 for the 7-d recall interview, r = 0.81 for the Godin-Shephard self-administered survey, and r = 0.93 for a simple activity rating. For the former two measures, reliability improved with age but was significant at all ages, and for the last measure there were no age effects. Memory skills and obesity status were not related to the reliability of recall, but males were more reliable reporters than females. Validity of the 7-d recall was determined by comparing heart rate monitoring records with recalls of very hard activities on the same day. A correlation of 0.53 (P < 0.001) for the total group supported the validity of the reports. Validity improved with age, but validity coefficients were significant in all age groups. These data indicate that physical activity recalls of children as young as the fifth grade are of adequate reliability and validity to use in research on physical activity in children.