Aminoglycoside nephrotoxicity has been reported in patients with sepsis, and several risk factors have been described. Once-daily dosing and shorter treatment have reduced nephrotoxicity risk, and simplified aminoglycoside monitoring. This review focuses on nephrotoxicity associated with aminoglycosides in the subset of patients with septic shock or severe sepsis. These patients are radically different from those with less severe sepsis. They may have, for instance, renal impairment due to the shock per se, sepsis-related acute kidney injury, frequent association with pre-existing risk factors for renal failure such as diabetes, dehydration and other nephrotoxic treatments. In this category of patients, these risk factors might modify substantially the benefit-risk ratio of aminoglycosides. In addition, aminoglycoside administration in critically ill patients with sepsis is complicated by an extreme inter- and intra-individual variability in drug pharmacokinetic/pharmacodynamic characteristics: the volume of distribution (Vd) is frequently increased while the elimination constant can be either increased or decreased. Consequently, and although its effect on nephrotoxicity has not been explored, a different administration schedule, i.e. a high-dose once daily (HDOD), and several therapeutic drug monitoring (TDM) options have been proposed in these patients. This review describes the historical perspective of these different options, including those applying to subsets of patients in which aminoglycoside administration is even more complex (obese intensive care unit [ICU] patients, patients needing continuous or discontinuous renal replacement therapy [CRRT/DRRT]). A simple linear dose adjustment according to aminoglycoside serum concentration can be classified as low-intensity TDM. Nomograms have also been proposed, based on the maximum (peak) plasma concentration (Cmax) objectives, weight and creatinine clearance. The Sawchuk and Zaske method (based on the determination of Cmax and an intermediate aminoglycoside assay before minimum plasma concentration) and the Bayesian method were both classified as high-intensity TDM programmes. Given the mean cost of aminoglycoside nephrotoxicity, these programmes may be cost-effective if its prevalence is above 10 %. However, none of these high-intensity TDM programmes have demonstrated a reduction of aminoglycoside-associated nephrotoxicity in patients with septic shock. Therefore, the questions remain as to, first, whether a TDM programme is relevant and, second, what intensity of TDM is achievable in the ICU, i.e. how it can be practically applied in the ICU setting where urgent care and high workload are substantial obstacles to a sophisticated, optimized aminoglycoside administration.