The safety of dietary protein and phosphorous restriction was evaluated in the Modification of Diet in Renal Disease (MDRD) Study. In Study A, 585 patients with a glomerular filtration rate (GFR) of 25 to 55 ml/min/1.73 m2 were randomly assigned to a usual-protein diet (1.3 g/kg/day) or a low-protein diet (0.58 g/kg/day). In Study B, 255 patients with a GFR of 13 to 24 ml/min/1.73 m2 were randomly assigned to the low-protein diet or a very-low-protein diet (0.28 g/kg/day), supplemented with a ketoacid-amino acid mixture (0.28 g/kg/day). The low-protein and very-low-protein diets were also low in phosphorus. Mean duration of follow-up was 2.2 years in both studies. Protein and energy intakes were lower in the low-protein and very-low-protein diet groups than in the usual-protein group. Two patients in Study B reached a "stop point" for malnutrition. There was no difference between randomized groups in the rates of death, first hospitalizations, or other "stop points" in either study. Mean values for various indices of nutritional status remained within the normal range during follow-up in each diet group. However, there were small but significant changes from baseline in some nutritional indices, and differences between the randomized groups in some of these changes. In the low-protein and very-low-protein diet groups, serum albumin rose, while serum transferrin, body wt, percent body fat, arm muscle area and urine creatinine excretion declined. Combining patients in both diet groups in each study, a lower achieved protein intake (from food and supplement) was not correlated with a higher rate of death, hospitalization or stop points, or with a progressive decline in any of the indices of nutritional status after controlling for baseline nutritional status and follow-up energy intake. These analyses suggest that the low-protein and very-low-protein diets used in the MDRD Study are safe for periods of two to three years. Nonetheless, both protein and energy intake declined and there were small but significant declines in various indices of nutritional status. These declines are of concern because of the adverse effect of protein calorie malnutrition in patients with end-stage renal disease. Physicians who prescribe low-protein diets must carefully monitor patients' protein and energy intake and nutritional status.