Background: Limited studies have evaluated the utility of scoring systems in the pediatric emergency department (PED) and no studies have evaluated their ability to predict hospital length of stay (LOS) and the usage of Observation units (OUs). Objective: To evaluate the utility of the Pediatric Early Warning Score (PEWS) in predicting LOS in pediatric patients and thus anticipate admission to an OU versus the pediatric ward. Methods: A retrospective study of pediatric inpatients (0 to 18 years) at an inner-city community hospital between January 2014 and December 2014. Patients with psychiatric illness, non-medical reasons for hospital stay, and those not discharged to 'home' were excluded. Demographic data, PEWS in the ED, and LOS for each patient were recorded and analyzed. Results: A total of 719 patients were analyzed. PEWS range was 0 to 8. The mean LOS was 56.8 hours for patients with PEWS 0-1 compared to 62.7 hours for patients with PEWS ≥2 (p=0.02). There was a significant difference in PEWS for LOS ≤24 and ≤36 hours in comparison to those with LOS >24 hours and >36 hours, respectively (p<0.001). Overall, the PEWS correlated with LOS (r=0.11, p=0.002). Age correlated inversely with LOS (r=-0.16, p<0.001), without correlation to PEWS (r=-0.002, p= 0.96). Conclusions: PEWS correlated weakly with LOS. A statistically significant lower PEWS was observed for patients who had short stays (both ≤24 and ≤36 hours) in comparison to those requiring longer inpatient care. Therefore, the PEWS is a useful tool to predict LOS and aid ED physicians to determine disposition, although further prospective studies in centers with OUs would better characterize its ability to suggest admission to an OU compared to the wards.