A prognostic study was retrospectively performed on a cohort of 201 ESRD patients having started CAPD in one single centre between 1983 and 1993 (mean follow-up 24 months). At the end of the study 74 patients had died, 63 were transplanted, and 19 had been transferred to another dialysis modality. The purpose of the study was to estimate patient and technique survivals according to clinical and biological variables recorded at the beginning of the treatment. The authors decided to focus on the possible prognostic role of urea kinetic parameters which are still controversial in the literature. Concerning patient survival, the Cox proportional hazards model selected seven prognostic factors, including urea Kt/V (relative risk = 1.69). The division of the study group according to three levels of Kt/V (< or = 1.7, between 1.7 and 2.2, and > 2.2) allowed us to estimate 1.7 as an index of an adequate CAPD prescription. NPCR at the limit of significance for patient survival was clearly a prognostic factor for technique survival (relative risk = 2.50). Creatinine clearance was closely linked to both technique and patient survival (relative risk > 3). These three factors, in addition to the clinical and biological variables generally described in the literature, must be taken into account in the evaluation of a prognosis for CAPD.