Claims-based utilization data collected in four Medicaid demonstration programs are used to examine the impact of primary care case management on patterns of reliance on the emergency room as a source of services. The experience of stratified random samples of AFDC (Aid to Families with Dependent Children) adults and children in the demonstration programs is compared with that of equivalent samples from comparison groups in traditional Medicaid programs. Data were analyzed in multivariate models controlling for person characteristics in a pre/post, demonstration/comparison design. Results indicated large reductions in the proportion of persons with at least one emergency room visit ranging from 27 to 37 per cent for children and 30 to 45 per cent for adults. Use levels for persons with at least one ER visit are less substantially affected. The findings suggest that gatekeeping designs can alter patterns of enrollee use with respect to the emergency room and some discussion of these implications is presented.