BACKGROUND : Levels of the biomarkers presepsin and procalcitonin are affected by renal function. We evaluated the accuracies of presepsin and procalcitonin levels for diagnosing sepsis in patients with and without acute kidney injury (AKI). METHODS : We evaluated patients with presepsin and procalcitonin data, and classified them into AKI and non-AKI groups based on the Kidney Disease Improving Global Outcomes criteria. Each group was then subdivided according to sepsis status for each stage of AKI. Receiver operating characteristic curve analyses were used to investigate the accuracies of biomarker levels for diagnosing sepsis. RESULTS : In the non-AKI group, the area under the curves (AUCs) for procalcitonin and presepsin levels were 0.897 and 0.880, respectively (p = .525) and optimal cut-off values were 0.10 ng/ml (sensitivity: 85.1%, specificity: 79.1%) and 240 pg/ml (sensitivity: 80.9%, specificity: 83.2%), respectively. In the stage 3 subgroup, the AUC for procalcitonin (0.946) was significantly higher than that for presepsin (0.768, p < .001). The optimal cut-off values for diagnosing sepsis were 4.07 ng/ml (sensitivity: 87.2%, specificity: 93.5%) for procalcitonin and 500 pg/ml (sensitivity: 89.7%, specificity: 59.7%) for presepsin. CONCLUSIONS : In patients with severe AKI, the accuracy of the diagnosis of sepsis with procalcitonin was significantly higher than with presepsin.