Modifications of antibiotic pharmacokinetic parameters have been reported in critically ill patients, resulting in a risk of treatment failure. We aimed to determine optimised amikacin (AMK), gentamicin (GEN) and tobramycin (TOB) intravenous dosing regimens in this patient population. Patients admitted to the medical ICU and treated with AMK, GEN or TOB were included. Analyses were performed using a parametric population approach. Monte Carlo simulations were performed and the probability of target attainment (PTA) was calculated using C-max/MIC >= 8 and trough concentrations as targets. A total of 117 critically ill hospitalised patients were studied. Median values (interindividual variability, omega(2)) of clearance were 3.51 (0.539), 3.53 (0.297), 2.70 (0.339) and 5.07 (0.339) L/h for AMK, GEN, TOB, and TOB in cystic fibrosis (CF), respectively. Median values (omega(2)) of central volume of distribution were 30.2 (0.215), 20.0 (0.109) and 25.6 (0.177) L for AMK, GEN and TOB, respectively. Simulations showed that doses should be adjusted to actual body weight and creatinine clearance (CLCR) for AMK and GEN, and according to CLCR and presence of CF for TOB. In conclusion, our recommendations for treating Pseudomonas aeruginosa infections in this population include using initial doses of 35 mg/kg for AMK or 10 mg/kg for TOB (CF and non-CF patients). GEN demonstrated the best rates of target attainment against Staphylococcus aureus infections with a dose of 5 mg/kg. As high aminoglycoside doses are required in this population, efficacy and safety targets are conflicting and therapeutic drug monitoring remains an important tool to manage this issue. (C) 2020 Elsevier Ltd and International Society of Antimicrobial Chemotherapy. All rights reserved.