Patients with chronic renal failure are commonly started on renal replacement therapy (RRT) as soon as (or, in some centers, before) the usual criteria for severity are met, i.e., GFR <10 ml/min for nondiabetic patients and <15 ml/min for diabetic patients. To determine whether RRT can safely be deferred beyond this point, adults with all types of chronic renal failure who met these criteria on presentation (23 patients) or who reached these levels of severity during treatment (53 patients) were managed conservatively until RRT was judged necessary by their chosen dialysis or transplantation team, without input into this decision from the present authors. Patients were prescribed a very low protein diet (0.3 g/kg) plus supplemental essential amino acids and/or ketoacids and followed closely. The intervals between the time at which GFR became less than 10 ml/min (15 ml/min in diabetic patients) and the date at which renal replacement therapy was started were used as estimates of renal survival on nutritional therapy. Kaplan-Meier analysis showed median renal survival of 353 d. Acidosis and hypercholesterolemia were both predictive of shorter renal survival. Signs of malnutrition did not develop. Final GFR averaged 5.6 +/- 1.9 ml/min. Two patients died; thus, annual mortality was only 2.5%. Hospitalizations totaled 19 in 93 patient-years of treatment, or 0.2 per year. Thus, these well motivated patients with GFR <10 ml/min (<15 ml/min in diabetic patients) were safely managed by diet and close follow-up for a median of nearly 1 yr without dialysis. It is concluded that further study of this approach is indicated.