BACKGROUND: To better measure physical activity (PA) in outdoor environments, McKenzie and colleagues developed the System for Observing Play and Recreation in Communities (SOPARC). However, previous SOPARC research has focused on adults, seniors, teens and children. One avenue for extending this work is to expand the child age group code to capture important nuances that can influence children's PA and their environments. This study reports on the reliability of a measure designed to account for PA in parks among children in different childhood age groups. METHODS: Three groups were developed: 0 to 5-years-old (Young Children); 6 to 12 (Middle Childhood) and 13 to 18 (Older Children) based on Erikson's stages of child development. Data were obtained by direct observation in 3 neighborhood parks in Raleigh, NC and 20 neighborhood parks in Durham, NC. RESULTS: Kappa coefficients showed high agreement for all age group, gender, and PA codes. For the 3 assessments, the results show that the 3 age group category exhibit acceptable reliability for measuring PA in parks among children. CONCLUSIONS: The reliability of measuring PA among children by segmenting children by 3 age groups was established. This approach is recommended for future studies of PA among children in parks and other outdoor environments.