Age/sex registries have been examined as a method of estimating the number of individuals served by a primary care practice. These data can be used in estimating disease frequency from primary care encounter data. The experience with age/sex registries in the Ambulatory Sentinel Practice Network (ASPN) has identified three major sources of error when registry data are used to estimate disease rates: (1) studies using medical encounter data exclude those individuals who do not seek medical care, (2) visitation is not random and is a function of variables in addition to disease incidence, and (3) encounter data from primary care practices are incomplete due to reporting problems and patient-initiated visits to other health care providers. Despite these limitations, age/sex registries can provide a practical tool for estimating disease rates in appropriate settings, assessing the generalizability of results, and assessing the feasibility of studies in practice based research. Further research about age/sex registries is needed to improve disease rate estimation as well as to better define methods. An age/sex registry enumerates a population by age and sex categories. Such a registry of the patients cared for by a medical practice represents a useful tool for practice based research. This paper briefly reviews the background of age/sex registries in North America, describes the experience of the Ambulatory Sentinel Practice Network with age/sex registries, identifies problems in using age/sex registries to provide denominator data for disease frequency estimation, and explores other uses for age/sex registry data.