Objective:Vasopressors are one of the main treatments for severe hypotension or shock, which commonly occurs in intensive care unit (ICU) patients. However, only a few studies have been conducted on the appropriate timing for vasopressor weaning. This study aims to explore the effect of blood pressure at vasopressor weaning on the probability of in-hospital mortality.Design: Single-center retrospective observational study.Setting: ICU from Beth Israel Deaconess Medical Center between 2008 and 2019.Patients: ICU patients who received vasopressor treatment were selected. Patients younger than 18 years old, died before vasopressor weaning or without blood pressure measurement at weaning were excluded. Finally, 8,298 patients were included.Result: General additive model (GAM) result showed that blood pressures at weaning had “U-shape” non-linear relationship with in-hospital mortality probability. The optimal levels of weaning mean arterial pressure(WMAP), weaning systolic blood pressure(WSBP), and weaning diastolic blood pressure(WDBP) were 85, 120, and 65 mmHg, respectively. Subgroup analysis showed the optimal WMAP, WSBP,WDBP has deviations between diagnoses. The “cut-point” of a lower mortality probability for WMAP was 65 mmHg. ROC curves showed that mean arterial blood pressure as an indicator exhibited the best prediction performance. Cox regression demonstrated that patients with WMAP equal to or higher than 65 mmHg will have 61% lower risk of in-hospital mortality.Conclusion: WMAP is a powerful indicator for in-hospital mortality, and its value should be greater than 65 mmHg and close to 85 mmHg to reach the highest survival probability for ICU patients after vasopressor treatment.