OBJECTIVE: Observation unit (OU) use has been promoted recently to decrease resource utilization and costs for select patients, but little is known about the operations of pediatric OUs. This study aimed to characterize the infrastructure and function of OUs within freestanding children's hospitals and to compare characteristics between hospitals with and without OUs. METHODS: All 43 freestanding children's hospitals that submit data to the Pediatric Health Information System were contacted in 2013 to identify OUs that admitted unscheduled patients from their emergency department (ED) in 2011. Semistructured interviews were conducted with representatives at hospitals with these OUs. Characteristics of hospitals with and without OUs were compared. RESULTS: Fourteen (33%) of 43 hospitals had an OU during 2011. Hospitals with OUs had more beds and more annual ED visits compared to those without OUs. Most OUs (65%) were located in the ED and had <12 beds (65%). Staffing models and patient populations differed between OUs. Nearly 60% were hybrid OUs, providing scheduled services. OUs lacked uniform outcome measures. Themes included: admissions were intuition based, certain patients were not well suited for OUs, OUs had rapid-turnover cultures, and the designation of observation status was arbitrary. Challenges included patient discontent with copayments and payer-driven utilization reviews. CONCLUSIONS: OUs were located in higher volume hospitals and varied by location, size, and staffing. Most functioned as hybrid OUs. OUs based admissions on intuition, had staffing cultures centered on rapid turnover of patient care, lacked consistent outcome measures, and faced challenges regarding utilization review and patient copayments.