In this longitudinal study in two prepaid group practices, many more children stayed at the same level of use of services over a six- to ten-year period than would be expected if use of services had distributed randomly. Overall, about 13% remained consistently in the highest third of the distribution of use, and another 13% remained consistently in the middle or lowest third. If use of services had distributed randomly, 4% and 7%, respectively, would have been in these groups. Conversely, many fewer children (25%) showed fluctuating patterns of use over time than would be expected by chance alone (37%). Although the reasons for this phenomenon are unknown (and may be multiple), the findings have implications both for clinical care and for development of policy regarding the organization and financing of health services for children.