Background: Early de-escalation of antibiotics in sepsis may be safe and effective. In our study, we performed a retrospective chart review of patients admitted to a large municipal hospital who were treated for severe sepsis or septic shock to compare outcomes in patients who experienced early de-escalation (DEG) with outcomes of those who did not (NDG).Methods: The observational study was conducted at Bellevue Hospital Center (an 850-bed municipal hospital affiliated with New York University School of Medicine, New York, NY). Patients admitted from January 1 to December 31, 2015, who were treated for severe sepsis or septic shock during any time of their hospital stay were reviewed for the study. De-escalation was defined as narrowing or discontinuation of 1 or more antimicrobial therapies < 3 days after sepsis onset. Results: Overall, 277 patients were included (DEG, 90 patients, 32%; NDG, 187 patients, 68%). The groups were similar in terms of sex, comorbidities, length of stay, and severity of illness: septic shock (47% DEG vs 49% NDG; P = .693) and ICU stay (27% DEG vs 32% NDG; P = .406). DEG patients were slightly older than NDG patients: (DEG age, 63+16 years vs NDE age, 58+16 years; P = .028). There was no difference in hospital mortality (8% DEG vs 12% NDE; P = .257). Nearly half of the patients in both groups (46% DEG and 47% NDG) had no causative microorganisms identified using conventional microbiology culture. The common sources of primary infection were respiratory, urinary tract, and gastrointestinal infections, and these were not different between groups. Also, 69% of DEG patients and 79% of NDG patients received antibiotics for >7 days (P = .002). Empiric intravenous vancomycin was initiated in 83% in DEG patients and 74% in NDG patients at sepsis diagnosis. Although organisms covered by intravenous vancomycin were isolated from only 17% of patients in DEG and 23% in NDG, vancomycin was continued for >5 days in 34% of DEG patients and 50.3% of NDG patients (P < .001). 60% of patients in DEG and 61% in NDG were seen by infectious diseases specialists (ID). Patients with infectious diseases consultations had significantly more comorbidities, were more frequently in the ICU, had higher MDRO isolation and longer hospital stays, but they were still de-escalated without a difference in mortality. Conclusions: Microbiology data did not contribute to early de-escalation of antibiotics in this study. This finding may be related to the high percentage of negative culture and unavailability of rapid molecular diagnostic tests. Shorter duration of antibiotics (including vancomycin) was not associated with worse outcome in these severely ill patients.Funding: NoneDisclosures: None