Acute kidney injury (AKI) is the new consensus term for acute renal failure. The term describes a continuum of kidney injury, a common condition in the critically ill and after major surgery, which is associated with increased mortality. The incidence of AKI in intensive care unit patients in Australia is >30% and sepsis is a major contributory factor. However, there is limited knowledge about its incidence after major surgery, except for cardiac surgery. The creation of staged AKI classification systems (RIFLE [Risk, Injury, Failure, Loss, End-stage], Acute Kidney Injury Network and the new Kidney Disease: Improving Global Outcomes criteria) has accelerated progress in critical care nephrology research by showing that even small changes in serum creatinine are associated with increased risk of death and that this risk increases progressively with severity of AKI. Recent thought and research has cast doubt over previously accepted pathophysiological views of AKI. Moreover, terms such as 'prerenal azotaemia' and 'acute tubular necrosis' are now being challenged as lacking validity, having little supportive evidence and carrying limited clinical utility. In this review, we explore the limitations of animal and human models of AKI and the implications of recent research on our current understanding of the pathophysiology of AKI. In addition, we describe conventional and novel diagnostic methods and therapies, and explore the clinical implications of the effect of fluid administration and perioperative management. Finally, we identify priorities for clinical investigations and future directions in AKI research.