Background. Accurate assessment of physical activity (PA) in children and adolescents is required to establish PA levels, monitor changes and inform public healthy policy. This study systematically reviews the literature to determine the extent of agreement between indirect (e.g., questionnaire) and direct (e.g., accelerometry) assessments of PA in pediatric populations (≤19 years). Methods. Literature was identified through searching electronic databases (e.g., MEDLINE, EMBASE), websites of relevant organizations and conference abstracts until April 2007. Studies were included if they collected indirect and direct measures of PA in pediatric populations and were reported in English. Quality of included studies was appraised using a modified Downs and Black tool. Results. A total of 83 studies were included; 24 describing comparable data and 59 including a correlation analysis. The majority of correlations reported between indirect and direct measures were low-to-moderate (range: −0.56–0.89). Overall, 72% of the indirect measures overestimated the directly measured values. Combined gender, as well as male- and female-only data comparing indirect measures to accelerometery, heart rate monitoring or direct observation, all reported an overestimation by indirect method. A similar trend was observed in combined gender data comparing indirect measures with doubly labelled water; however, the opposite trend was observed in the male- and female-only data with a slight underestimation by indirect measure. Conclusions. Substantial discrepancies and moderate correlations between indirect methods and direct measures of assessing PA in pediatric populations are of concern, especially when trying to establish relationships with health outcomes.