Objective <p>This study aimed to investigate the association between the Prognostic Nutritional Index (PNI) and 90-day mortality among patients with sepsis.</p> Methods <p>We conducted a retrospective cohort study utilizing data from the MIMIC-IV database for patients diagnosed with sepsis. Covariates were selected using Cox regression models. The association between PNI and 90-day mortality was assessed using multivariable Cox proportional hazards models. We further examined the potential nonlinear relationship using restricted cubic spline(RCS) analysis. Kaplan-Meier survival curves were plotted for these groups, and the differences in 90-day survival rates were compared using the log-rank test.</p> Results <p>A total of 5,786 septic patients were included in the final analysis. The overall 90-day mortality rate after ICU admission was 35.0% (<i>n</i> = 2,027). In the fully adjusted Cox model, a higher PNI was associated with a reduced risk of 90-day mortality. Specifically, for each one-unit increase in PNI, the risk of 90-day mortality decreased by 1% (adjusted HR = 0.99, 95% CI: 0.98 ~ 1.00, <i>P</i> = 0.004). The RCS analysis revealed a significant nonlinear relationship between PNI and 90-day mortality (<i>P</i>_nonlinearity &lt; 0.001). The log-rank test confirmed a statistically significant difference in survival curves between the low and high PNI groups (<i>P</i> &lt; 0.0001).</p> Conclusion <p>PNI demonstrates a significant nonlinear inverse association with 90-day mortality in patients with sepsis. These findings suggest that PNI could serve as a valuable prognostic tool. Clinicians may use PNI to identify high-risk patients and implement targeted interventions to potentially improve clinical outcomes.</p>