This paper describes a new ambulatory case-mix system developed at The Johns Hopkins University and known as Ambulatory Care Groups (ACGs). ACGs categorize a person into one of 51 categories based on the diseases and conditions for which they received treatment over a period of time, such as a year. ACGs can be used to describe the "illness-burden" of a population and are up to ten times more predictive of ambulatory care resource use than age and sex alone. ACGs can be determined using a computerized "grouper" software package based on ICD-9-CM diagnosis codes and demographic information presently found in virtually all claims or encounter data systems. They were developed and tested at four HMOs and a state's Medicaid program. This paper discusses the potential application of ACGs to analysis, financing, and management of ambulatory care, specifically as it relates to utilization review (UR), quality assurance (QA) and the adjustment of capitation payment within managed care settings.