Rates of progression of chronic renal failure were compared in patients receiving alternately an amino acid supplement (AA) and a ketoacid supplement (KA) to a very low protein (0.3 g/kg), low phosphorus (7 to 9 mg/kg) diet. The first supplement was randomly chosen. Bias due to carryover effects was minimized by delaying the regression analysis until one month after starting or changing supplements. In order to minimize possible bias caused by initiating the two supplements at differing levels of severity, a multiple crossover design was used (ABA, BAB, ABAB, or BABA) with at least four GFR's in each treatment period (except for three GFR's in one instance). Sixteen patients completed the protocol; five dropped out. Average starting GFR's were nearly identical for the two supplements (15.4 and 15.9 ml/min). For each patient, mean progression on KA was compared with mean progression on AA. Thirteen out of 16 patients progressed more slowly on KA than AA. On the average, progression on KA was significantly slower (95% confidence limits = -0.36 to 0.09 ml/min/month) than on AA (-0.91 to -0.41 ml/min/month; P = 0.024). There was no significant difference in estimated protein intake, phosphate excretion, or mean arterial pressure between KA and AA periods. Serum triglyceride concentration was significantly lower on KA (P = 0.0026). 17-hydroxycorticosteroid excretion was also lower (P = 0.031). We conclude that KA slow progression, relative to AA, independently of protein or phosphorus intake, in patients on this regimen.