Objective: Early initiation of renal replacement therapy (RRT) effect on survival and renal recovery of critically ill patients is still uncertain. We aimed to systematically review current evidence comparing outcomes of early versus late initiation of RRT in critically ill patients. Methods: We searched the Medline (via Pubmed), LILACS, Science Direct, and CENTRAL databases from inception until November 2016 for randomized clinical trials (RCTs) or observational studies comparing early versus late initiation of RRT in critically ill patients. The primary outcome was mortality. Duration of mechanical ventilation, intensive care unit (ICU) length of stay (LOS), hospital LOS, and renal function recovery were secondary outcomes. Meta-analysis and trial sequential analysis (TSA) were used for the primary outcome. Results: Sixty-two studies were retrieved and analyzed, including 11 RCTs. There was no difference in mortality between early and late initiation of RRT among RCTs (odds ratio [OR] = 0.78; 95% confidence interval [CI]: 0.52-1.19; I² = 63.1%). Trial sequential analysis of mortality across all RCTs achieved futility boundaries at both 1% and 5% type I error rates, although a subgroup analysis of studies including only acute kidney injury patients was not conclusive. There was also no difference in time on mechanical ventilation, ICU and hospital LOS, or renal recovery among studies. Early initiation of RRT was associated with reduced mortality among prospective (OR = 0.69; 95% CI: 0.49-0.96; I² = 85.9%) and retrospective (OR = 0.61; 95% CI: 0.41-0.92; I² = 90.9%) observational studies, both with substantial heterogeneity. However, subgroup analysis excluding low-quality observational studies did not achieve statistical significance. Conclusion: Pooled analysis of randomized trials indicates early initiation of RRT is not associated with lower mortality rates. The potential benefit of reduced mortality associated with early initiation of RRT was limited to low-quality observational studies.