Urea kinetic modelling was performed serially, over 24 months, on 55 patients undergoing hemodialysis and eight patients receiving peritoneal dialysis. The data obtained, together with changes in therapy aimed at increasing or decreasing the normalized dose of dialysis [KT/V (urea)], suggested the dependence of dietary protein intake and protein catabolic rate (PCR; g/kg/d) on the KT/V (urea). The studies also indicated that the nature of this relationship may be dependent upon the dialysis treatment used; dialysis by AN69S membrane hemodialyzers required less KT/V (urea) than hemodialysis by cellulosic membranes to obtain a given PCR. This difference may be explained by the beneficial effect of removal of "middle molecular weight" uremic toxins by the AN69S membrane, which has a different solute clearance profile than the cellulosic membrane. The studies also indicated a similar relationship between PCR and KT/V (urea) for peritoneal dialysis. With this form of therapy, however, it is difficult to obtain a PCR greater than 1 g/kg/d without first achieving very high values for KT/V (urea). It is postulated that this is due to an independent adverse effect of peritoneal dialysate in suppressing appetite. The data presented suggest that the conclusions of the National Cooperative Dialysis Study may be reinterpreted by assigning a major role to the nutritional status of patients in morbidity, with satisfactory nutritional status attained only in patients receiving adequate dialysis which, in turn, ensures control of plasma urea levels. Studies to prove this hypothesis are indicated.