Presepsin (PSP) is a viable biomarker for the detection of bacterial infection, but it lacks accuracy when acute kidney injury (AKI) develops. Herein, we evaluated the diagnostic and prognostic value of PSP in predicting postoperative sepsis after abdominal surgery respective to the degree of AKI. A total of 311 patients who underwent abdominal surgery and were admitted to a surgical intensive care unit were enrolled and classified into non-AKI, mild-AKI (stage 1, stage 2 and stage 3 without renal replacement therapy (RRT)) and severe-AKI (stage 3 with RRT) group, according to the Kidney Disease Improving Global Outcomes criteria. In each group, PSP and other biomarkers were statistically analyzed between non-sepsis and postoperative sepsis at the admission (T0), 24 h (T1), 48 h (T2) and 72 h (T3) after surgery. In non-AKI and mild-AKI group, PSP levels were significantly higher in postoperative sepsis than non-sepsis group, whereas no difference was detected in the severe-AKI group. Cutoff values of PSP in the mild-AKI group for the prediction of postoperative sepsis were 544 pg/mL (AUC: 0.757, p < 0.001) at T0 and 458.5 pg/mL (AUC: 0.743, p < 0.001) at T1, significantly higher than in non-AKI group. In multivariate analysis, predictors of postoperative sepsis in the mild-AKI group were PSP at T2 (odds ratio (OR): 1.002, p = 0.044) and PSP at T3 (OR: 1.001, p = 0.049). PSP can be useful for predicting newly developed sepsis in patients with transient AKI after abdominal surgery with modified cutoff values.