Acute kidney injury (AKI) is now well recognized as an independent risk factor for increased morbidity and mortality particularly when dialysis is needed. Although renal replacement therapy (RRT) has been used in AKI for more than five decades, there is no standard methodology to predict which AKI patients will need dialysis and who will recover renal function without requiring dialysis. The lack of consensus on what parameters should guide the decision to start dialysis has led to a wide variation in dialysis utilization. A contributing factor is the lack of studies in the modern era evaluating the relationship of timing of dialysis initiation and outcomes. Although listed as one of the top priorities in research on AKI, timing of dialysis initiation has not been included as a factor in large, randomized controlled trials in this area. In this review we will discuss the criteria that have been used to define early vs. late initiation in previous studies on dialysis initiation. In addition, we propose a patient-centered approach to define early and late initiation that could serve as framework for managing patients and for future studies in this area.