An 8-year experience on CAPD, in a single center with all treatments of ESRF (end-stage renal failure) available, is presented. Method choice was left to the patient, after extensive counselling. However, CAPD selection was very negative, and CAPD patients were older, with a much larger percentage of diabetics and loaded by more risk factors, suggesting an influence of the staff preferences on patient choice. After a first period with unsatisfactory results, we obtained an important improvement of patient and method survival coinciding with the introduction of a new connector with disinfectant (Y-system) which allowed a reduction of peritonitis rate to 1 episode for 36 patient/months. For the period 1.1.81 to 31.12.86 a comparison was made (life table analysis) between new ESRF patients placed initially on CAPD or on HD. The 5-year survival was not statistically different in spite of the very negative CAPD selection of patients, who were 10 years older, on the average. Excluding diabetics, survival curves were identical in the two methods. Age at death and causes of death were not different. Method survival was better on HD (98% vs. 71% on CAPD, at 5 years, p less than 0.01): significance and limits of this evaluation are discussed. Drop-out figures were definitely lower than in the literature and this was attributed to the sharp reduction in peritonitis rate. Only 1.7% of CAPD patients discontinued the method due to inadequate ultrafiltration. In 29 CAPD and 28 HD patients with more than 4 years treatment some biochemical and clinical data were compared. Serum cholesterol was significantly higher and serum proteins lower in CAPD.(ABSTRACT TRUNCATED AT 250 WORDS)