Sepsis remains associated with a high mortality rate despite recent advances in treatment. Traditional biomarkers are inadequate for stratification of patients by sepsis severity. We examined use of the baseline concentration of fibroblast growth factor 19 (FGF19) in predicting 28-day mortality from sepsis.A total of 220 consecutive adult patients with sepsis who were admitted to our intensive care unit (ICU) during 2020 were prospectively recruited. Patients were categorized as survivors or non-survivors according to status at 28 days. Baseline concentrations of FGF19 and other parameters were measured. Receiver operating characteristic (ROC) analysis was used to determine the sensitivity, specificity, predictive value, and optimal cutoff of FGF19 in prediction of survival. Prognostic factors were identified using Cox regression analysis.The serum FGF19 concentration was much higher in non-survivors than in survivors (355.0 pg/ml [range: 37.2, 2315.6] vs. 127.3 pg/ml [5.7, 944.1]; P < 0.05]. ROC analysis indicated an FGF19 concentration of 180 pg/ml was the optimal cutoff value. Multivariable Cox regression analysis showed that FGF19 concentration and the change in sequential organ failure assessment (ΔSOFA) score at baseline were independently and significantly associated with 28-day mortality. ROC analysis indicated that FGF19 had a better predictive value than PCT or CRP. Although ΔSOFA had a better predictive value than FGF19, ΔSOFA and FGF19 together had a significantly better predictive value than ΔSOFA alone.Sepsis patients with high serum concentrations of FGF19 at ICU admission were associated with an increased risk of 28-day mortality in our ICU.Copyright © 2021. Published by Elsevier B.V.