OBJECTIVES The 23-hour observation units (OUs) may be used to avoid unnecessary hospital admissions. However, unexpected hospitalizations from the 23-hour OUs involve transfer of care and may decrease the efficiency and safety of care of the patient and the unit itself. The primary objective of this study was to determine the predictors of unexpected hospitalization for admissions to a pediatric 23-hour OU. METHODS This is an observational prospective cohort study of patients admitted to a pediatric 23-hour OU. Bivariate and multivariate regression analyses identify factors associated with unexpected hospitalization. RESULTS There were 4453 patients admitted to the 23-hour OU during the study. The overall rate of unexpected hospitalization was 20.3%; the mean 23-hour OU stay was 15 hours. Age, sex, race/ethnicity, and insurance status were not associated with increased unexpected hospitalization rates. Multivariate regression modeling revealed that unexpected hospitalization was associated with subgroups of resources used (intravenous medications and fluids, cardiorespiratory monitoring, respiratory therapist use, and supplemental oxygen), of subspecialty consultation, and of diagnosis categories (including asthma, adenitis, cellulitis, bronchiolitis, and esophageal foreign body ingestions). Experience of the health care provider involved in the care of the patient was not associated with increased unexpected hospitalization. CONCLUSIONS Most of the patients (80%) were successfully discharged from the 23-hour OU. Demographics of the patient and practitioner characteristics did not influence the risk of unexpected hospitalizations; however, certain patient diagnoses, use of resources,and subspecialty consultation did increase the risk of unexpected hospitalization and, therefore, may guide future admission criteria for pediatric 23-hour OU.