OBJECTIVES: To characterize hospital variation in use of observation stays to manage pulmonary embolism (PE) and its association with subsequent outcomes. METHODS: We performed a cross-sectional study of hospitals reporting ≥75 PE encounters of any type (emergency department, observation stay or inpatient admission) using data from the Premier Database between 11/2012-3/2015.  We included all hospital encounters for adults with a primary diagnosis of PE (415.1x), ≥1 diagnostic test claim for PE on day 0-2 and evidence of PE treatment.  Hospitals were divided into tertiles (Ts) based on the proportion of all PE encounters managed as an observation stay.  The association between observation stay utilization and the proportion of all PE encounters resulting in re-admission within the same or subsequent 2-months were compared across Ts using χ2 tests. RESULTS: Observation PE management increased over the study period (1.9%-5.4%; Pearson’s r=0.88, p<0.001). Of all hospitals reporting ≥1 PE encounter, 255 had ≥75 encounters (representing a total of 38,172 PE encounters) and were included in the analysis. Individual hospital use of observation for PE management varied from 0%-33.9%.  Mean hospital rates of PE observation stay by T were T1=0.1%, T2=2.2% and T3=7.9%.  Hospitals that used observation stays most frequently (T3) were more likely in the South or Mid-west (p<0.001), were a teaching hospital (p=0.03) and less likely to serve an urban population (p=0.02).  Hospitals in T3 (n=11,780 encounters) did not have an increased risk for all-cause (4.7% vs. 5.1%-5.4%, p=0.02), venous thromboembolism-(1.4% vs. 1.8%-2.0%, p=0.004) or major bleeding (0.3% vs. 0.2-0.3%, p=0.39)-related re-admission in the same or subsequent 2-months compared to T1 (n=12,940 encounters) and T2 (n=13,452 encounters). CONCLUSIONS:   PE management via observation stays increased significantly over recent years.  Hospitals more frequently utilizing observation stays did not experience more negative outcomes, such as re-admission.