Management of fluid overload is one of the most challenging problems in the care of critically ill patients with oliguric acute kidney injury. Various clinical practice guidelines support fluid removal using ultrafiltration during kidney replacement therapy. However, ultrafiltration is associated with considerable risks. Emerging evidence from observational studies suggests that both slow and fast rates of net fluid removal (that is, net ultrafiltration (UFNET)) during continuous kidney replacement therapy are associated with increased mortality compared with moderate UFNET rates. In addition, fast UFNET rates are associated with an increased risk of cardiac arrhythmias. Experimental studies in patients with kidney failure who were treated with intermittent haemodialysis suggest that fast UFNET rates are also associated with ischaemic injury to the heart, brain, kidney and gut. The UFNET rate should be prescribed based on patient body weight in millilitres per kilogramme per hour with close monitoring of patient haemodynamics and fluid balance. Dialysate cooling and sodium modelling may prevent haemodynamic instability and facilitate large volumes of fluid removal in patients with kidney failure who are treated with intermittent haemodialysis, but the effects of this strategy on organ injury are less well studied in critically ill patients treated with continuous kidney replacement therapy. Randomized trials are required to examine whether moderate UFNET rates are associated with a reduced risk of haemodynamic instability, organ injury and improved outcomes in critically ill patients.