Despite the availability of clinical guidelines for the timing of dialysis initiation in both the United States and Canada, patients continue to start dialysis at very low levels of predicted glomerular filtration rate (GFR). A cross-sectional study was performed to determine the demographic and clinical characteristics of patients who started hemodialysis, their level of GFR, and mortality at 1 and 2 years following the initiation of dialysis. Retrospective data were collected on all eligible patients who commenced chronic hemodialysis in 1 tertiary care center in Canada from March 2001 to February 2005. Only those patients who had been followed by a nephrologist in the chronic kidney disease clinic before dialysis initiation were included (n=271). Seventeen percent of patients started hemodialysis late (GFR<5 mL/min/1.73 m(2)). Compared with the group of patients who started dialysis earlier, the late start group were significantly younger (p=0.008), had more females (p=0.013), more employed (p=0.051), less cardiac (p<0.001), and peripheral vascular disease (p=0.031), and were taking medication for hypertension (p=0.041). Serum albumin was lower in the late start group (p=0.023). At year 1, there was no difference in mortality rate while at year 2, the earlier the dialysis, the greater the mortality rate (p=0.022). After adjustment for demographic variables and comorbidities, only antihypertensive use had an independent but weak association with the 2 year mortality. Adjustment for all these variables eliminated the significant association noted for the 2 year mortality in the early versus late dialysis start. The survival benefit for late versus early dialysis start appears to be multifactorial and relates to a preponderance of clinical and demographic factors favoring a lengthened survival occurring in the late dialysis group. Our survival benefit findings suggest the premorbid health condition is a more important determinant of 2 year survival than the timing of dialysis initiation.