This study describes the distribution and length of stay (LOS) for children and adolescents in inpatient psychiatric facilities and identifies factors associated with LOS. The data are from the 1986 Client/Patient Sample Survey, a nationally representative survey conducted by the National Institute of Mental Health. The median national LOS, weighted to population levels, is more than two months for children compared with approximately two and one-half weeks for adolescents. The most important determinants of LOS are type of facility, age, insurance coverage, ethnicity, treatment history, and diagnosis. Key words: adolescents; children; diagnosis; Hispanic ethnic origin; inpatient psychiatric hospitals; insurance; length of stay Recent studies of trends in psychiatric services for children and adolescents reveal that admissions to psychiatric hospitals have increased during the past two decades (Burns, 1991; Pottick, Hansell, Gaboda, & Gutterman, 1993) at the same time that average length of stay (LOS) has decreased (Borchardt & Garfinkel, 1991). Possible reasons for shorter LOS are increased financial pressures and the recently developed consensus that for children and adolescents hospitalization should be used only as a last resort when children have severe psychiatric problems or when home and community-based services are inadequate (Stroul & Friedman, 1988). Ideally, we need to discover the optimum LOS for children and adolescents with specific psychiatric problems and service needs and then tailor inpatient programs to provide services in the context of comprehensive home and community-based outpatient programs. Of course, the present reality falls short of this ideal, and it is likely that LOS is influenced by a number of factors in addition to patient needs. The study discussed in this article examined the effects of several patient, insurance, and hospital factors on LOS. It built on our previous work, which found that insurance, illness, and age factors were associated with the selection of children and adolescents into inpatient or outpatient treatment (Pottick, Hansell, Gutterman, & White, 1995). With the same national data set from that study, we examined factors associated with LOS. We expected that the factors that influence pathways into care are different from factors that keep children in care. The analysis is based on the Andersen and Newman (1973) and Andersen (1995) model of health care utilization, for which the use of health care services was predicted by illness (or need) factors, predisposing factors, and enabling factors. Illness factors refer to indicators of illness that signal a need for services, such as illness diagnosis and severity. Predisposing factors refer to pre-existing characteristics of individuals that influence the likelihood of service use, such as demographic characteristics, illness history, and beliefs about health and services. Enabling factors refer to family and community resources that make utilization possible, including family resources and income, level of insurance coverage, availability of an adult to provide care in the home, and access to health facilities and providers in the community. The model's efficacy has been supported in part by a number of studies on physical health utilization, especially among adults. Recently, three studies (Padgett et al., 1993; Patrick et al., 1993; Pottick, Hansell, Gaboda, & Gutterman, 1995) found this model useful for explaining mental health utilization for children and adolescents. The present study used youth data from the 1986 Client/Patient Sample Survey (CPSS), a nationally representative survey conducted by the National Institute for Mental Health (NIMH) that describes the distribution and LOS of children and adolescents in different types of inpatient facilities in the United States. Earlier analyses of these data (for example, Burns, 1991; Pottick et al., 1995) compared the illness, predisposing and enabling characteristics of children and adolescents in inpatient, outpatient, and partial care. …