Recent guidelines from the US National Kidney Foundation Dialysis Outcomes Quality Initiative recommend an earlier start of dialysis treatment than has been common practice. Their implementation would have a substantial effect on patients' daily lives and would increase costs. The guidelines are largely opinion-based, because evidence is still lacking. As part of a prospective multicentre study in the Netherlands, we included, between January, 1997, and May, 1999, all new patients with end-stage renal disease, for whom data were available on residual renal function 0-4 weeks before the start of dialysis. We recorded date of death or censoring until August, 2000. 94 (37%) of 253 patients started dialysis treatment later than recommended by the US guideline. There was an increased mortality risk for these patients compared with those who started dialysis on time, although it was not significant (adjusted hazard ratio 1.66 [95% CI 0.95-2.89]). The adjusted difference in estimated survival time after 3 years on dialysis treatment was 2.5 months (1.1-4.0) in favour of timely starters. Conversely, the average delay in dialysis initiation for late starters, the extra time free of dialysis, was at least 4.1 months. Although we observed a gain in survival time with a timely start of dialysis, it is probably a reflection of initiating dialysis earlier in the disease. We question the benefit of putting this guideline into daily practice, given the current clinical evidence and the effects it would have on patients and dialysis resources.