This article describes a new case-mix methodology applicable primarily to the ambulatory care sector. The Ambulatory Care Group (ACG) system provides a conceptually simple, statistically valid, and clinically relevant measure useful in predicting the utilization of ambulatory health services within a particular population group. ACGs are based on a person's demographic characteristics and their pattern of disease over an extended period of time, such as a year. Specifically, the ACG system is driven by a person's age, sex, and ICD-9-CM diagnoses assigned during patient-provider encounters; it does not require any special data beyond those collected routinely by insurance claims systems or encounter forms. The categorization scheme does not depend on the presence of specific diagnoses that may change over time; rather it is based on broad clusters of diagnoses and conditions. The presence or absence of each disease cluster, along with age and sex, are used to classify a person into one of 51 ACG categories. The ACG system has been developed and tested using computerized encounter and claims data from more than 160,000 continuous enrollees at four large HMOs and a state's Medicaid program. The ACG system can explain more than 50% of the variance in ambulatory resource use if used retrospectively and more than 20% if applied prospectively. This compares with 6% when age and sex alone are used. In addition to describing ACG development and validation, this article also explores some potential applications of the system for provider payment, quality assurance, utilization review, and health services research, particularly as it relates to capitated settings.