Whether to use peritoneal dialysis (PD) or hemodialysis (HD) is a major decision in terms of clinical outcome and management implications; the final choice is difficult because of the conflicting results of comparisons reported in the literature. A review of studies comparing survival shows either superiority of HD, or superiority of PD, or equivalence of the two techniques, but an analysis of the comparisons as a whole brings to light two clear phases in the survival curves. In the first, residual renal function (RRF) gives PD an advantage, or at least puts it on the same level as HD. In the second phase, the reduction in Kt/V as RRF declines gives PD a potential risk. After a few years of PD treatment a sharp watch is therefore necessary to detect signs of under-dialysis promptly and to shift the patient to HD. In patients without RRF it is more difficult to control hypertension with PD and they are more prone to hyperhydration. Despite a widespread belief in the Eighties that PD was the treatment modality of election for diabetics, HD is in fact preferable in these patients, except younger ones. High-turnover and low-turnover bone lesions are more frequent respectively in HD and PD patients. Anemia is better controlled with PD. Blood lipids and nutritional indices are less well controlled with PD. Despite poor technical survival, the "pool" of patients treated with PD frequently reaches 20-30% because it is indicated as first treatment in a large proportion. PD preserves renal function better than HD and is useful while awaiting renal transplantation, with faster postoperative restoration of diuresis. The quality of life with PD as home treatment is usually better than with HD. In conclusion, dialytic centers should establish an integrated PD/HD programme as the two methods are not competitive but are different tools for the treatment and rehabilitation of uremic patients.