<p>Estimated glomerular filtration rate (eGFR) based on creatinine (eGFR_Creatinine) or cystatin C (eGFR_CystatinC) require steady-state conditions and thus have limitations in intensive care unit (ICU) patients. Gentamicin is a potential exogenous marker for eGFR but poorly investigated. This retrospective study included adult ICU patients (<i>≥</i> 18 years) treated with gentamicin and not on renal replacement therapy (RRT) at admission. eGFR_Creatinine and eGFR_CystatinC were calculated using the LM-rev and CAPA equations, respectively. Gentamicin clearance was estimated using a population pharmacokinetic model and used as eGFR_Gentamicin. Agreement between eGFRs vs. eGFR_Gentamicin and prediction of RRT and mortality for each eGFR were assessed. 254 patients were included of whom 11% (<i>n</i> = 28) received RRT later and 19% (<i>n</i> = 49) were dead at 30 days. The bias was 12 mL/min/1.73 m² and 8 mL/min/1.73 m², respectively, and the limits of agreement − 31–55 mL/min/1.73m² and − 46–62 mL/min/1.73m² for the agreement between eGFR_Gentamicin vs. eGFR_Creatinine, and for eGFR_Gentamicin vs. eGFR_CystatinC, respectively. The c-indexes for predicting RRT during ICU stay were 0.75 (0.64–0.86), 0.77 (0.66–0.88) and 0.80 (0.69–0.90) for eGFR_Creatinine, eGFR_CystatinC and eGFR_Gentamicin respectively, and for 30-day mortality 0.61 (0.52–0.70), 0.61 (0.52–0.70) and 0.63 (0.54–0.72) respectively. In ICU patients already receiving gentamicin, eGFR_Gentamicin derived from population PK models can be used to assess renal function and could potentially help improve dosing of other renally cleared drugs like the β-lactams during early phase of infections in the ICU.</p>