Hypoalbuminemia is the most powerful predictor of mortality in end-stage renal disease. Since protein-calorie malnutrition can decrease albumin synthesis it is assumed that hypoalbuminemia results principally from malnutrition in these patients, but albumin synthesis may also be decreased as part of the acute-phase response, and hypoalbuminemia can also result from redistribution of albumin pools or from albumin losses, We measured albumin synthesis, fractional catabolic rate, and distribution from the turnover of [I-125] human albumin in six hemodialysis patients with plasma albumin less than 35 mg/ml and in six patients with plasma albumin greater than 40 mg/ml. Patients with liver disease, HIV, or other infection were excluded. Both groups were maintained with high-flux polysulfone dialyzers for more than three months. Kt/V-urea and PCR were measured during each dialysis (N = 12 to 18/patient). A four-day calorie and protein intake was determined by dietary history and long-term nutritional status was determined anthropometrically. Measured variables included serum urea, creatinine, transferrin, and the positive acute-phase proteins alpha(2)-macroglobulin, C-reactive protein, ferritin, and IGF-1. Albumin synthesis was significantly reduced in the low albumin group. There were no differences in dietary intake, body composition, PCR, BUN, creatinine, or Kt/V-urea. Plasma albumin concentration correlated negatively with ferritin, C-reactive protein and alpha(2)-macroglobulin. Albumin synthesis rate correlated negatively with both alpha(2)-macroglobulin and Kt/V-urea. Both plasma albumin concentration and synthesis rate correlated positively with IGF-1, and both were independent of PCR and all other nutrition-related variables. Hypoalbuminemia was due to decreased albumin synthesis. The data suggest that albumin concentration and synthesis were primarily determined by non-nutritional factors in this well-dialyzed population, and that both were reduced partially as part of the acute-phase response.