As Community Support Programs for the chronically mentally ill expand, it becomes increasingly important to determine the number of individuals qualifying for these services. Although national prevalence data are currently available, they have only limited usefulness for program planning at state and local levels. Given the distinctive circumstances affecting each community's chronically ill population, their number, and their need for services, the authors propose identifying a patient cohort that approximates the local prevalence of persons needing Community Support Programs by using one of three methods. The first method identifies persom who have been hospitalized previously and who currently require outpatient psychiatric care. The second identifies persons previously hospitalized who require another hospitalization during a specified period of time. And, the third method identifies persons who are currently in outpatient treatment with a diagnosis of schizophrenia. Choice of method depends on definition of chronic mental illness, type of data available, local treatment philosophies, and health care system structure.