We examined health care utilization, clinical profiles (such as sociodemographic features, clinical severity), and outcomes (inpatient admission, revisit within 24 hours of discharge) of patients who were admitted to a 23-hour observation unit within the emergency service of a tertiary psychiatric hospital and hypothesized that a specific clinical profile (greater clinical severity, lower psychosocial functioning) predicted subsequent inpatient hospitalization. The medical records of all patients admitted to the observation unit from February 5, 2007, to February 4, 2012 (N = 2,158) were assessed for relevant data. Clinical severity and level of psychosocial functioning were assessed using Clinical Global Impressions-Severity (CGI-S) and Global Assessment of Functioning (GAF) scales, respectively. Overall, the patients seen were predominantly Chinese males > 36 years old who had diagnoses including stress-related, anxiety, affective spectrum, and psychotic disorders. The clinical severity score (CGI-S) improved significantly following discharge from the observation unit (t 1,1848 = 23.316; P < .001). Logistic regression analyses revealed that self-referred (P = .001), older patients (P = .007) with past psychiatric history (P = .019), lower GAF scores (P = .025), and less improvement of CGI-S scores (P = .001) were associated with inpatient admission after a 23-hour stay in the observation unit. Our study findings affirmed our hypothesis and supported the utility of the observation unit in monitoring the overall clinical status of patients, which was linked with subsequent inpatient admissions. Better management of these patients at the outpatient level can potentially decrease unnecessary hospitalization and reduce health care cost as well as illness burden on patients and caregivers.