On 578 unselected new patients followed from 1981 through 1993, 51% on CAPD and 49% on HD, long-term patient and method survivals, cause of death, and drop-out in the two methods were compared. Survival, adjusted for patient selection biases, was not different on CAPD and HD up to 10 years. 50% of the patients were still in their first treatment after 3.5 years on CAPD and after 7 years on HD, and 5 and 28% respectively, after 10 years. Patient survival on CAPD was not falsely improved by drop-outs. Drop-out is increasing for CAPD, mainly due to patient/partner burn-out, which should be relieved by a more liberal application of automated PD. Malnutrition is more frequent on CAPD than on HD but not for the elderly. In a 3 year prospective study on 60 CAPD and 34 HD patients serum albumin, nPCR and nutritional status, as assessed by SGA did not influence survival in each modality. Survival was similar with K(p,r)t/V > or = 1.7/week on CAPD and Kt/V > or = 1/treatment on HD, and worse below these values. On CAPD, a Kp,rt/V > or = 1.96 gave better survivals.