The objective was to review evidence addressing the optimal time to initiate dialysis treatment. The database was derived from an evidence-based review of the medical literature and from the Canada-United States peritoneal dialysis study. The publications were divided into (1) those addressing the clinical impact of early versus late referral to a dialysis program; (2) those evaluating the association between residual renal function at initiation of dialysis and the concurrent nutritional status; (3) those evaluating the association between residual renal function at initiation of dialysis and subsequent clinical outcomes, including patient survival. There were five studies evaluating early versus late referral, three cohort design and two case-control design. Late referrals had worse outcomes than early referrals. The former had more serious comorbidity and many had been noncompliant with follow-up. The latter were more likely to have hereditary renal disease. Renal function was slightly worse at initiation among those referred late. Three studies addressed the association between renal function at initiation of dialysis and concurrent nutritional status. Two showed decreased protein intake with diminished glomerular filtration rate (GFR). Poor nutritional status is associated with decreased patient survival among both incident and prevalent dialysis patients. The third study reported excellent patient survival among patients with late initiation of dialysis. These patients had received a supplemented low-protein diet and were not malnourished at initiation of dialysis. Three groups have studied the association between GFR at initiation of dialysis and clinical outcomes. Decreased GFR at initiation of dialysis is associated with a increased probability of hospitalization and death. None of these studies has used the rigorous randomized clinical trial design, and they are therefore subject to bias. Referral time bias, comorbidity, patient compliance, and starting time bias are potential confounders. A randomized clinical trial is required to resolve this important issue. However, there is sufficient evidence to justify initiation of dialysis at a Ccr of 9 to 14 mL/min if there is any clinical or laboratory evidence of malnutrition.