Albumin-synthesis rates were measured in nine patients with stable cirrhosis and compared with those of eight healthy volunteers by means of a new technique using stable isotopes. Four grams of L-[ l-lsC]leucine was injected over 10 min, and blood samples were drawn at intervals. Serum free [*3C]leucine enrichment, taken to be the precursor for albumin synthesis, and lSC enrichment of leucine in albumin, isolated with differential solubility in absolute ethanol from trichloroacetic acid-precipitated serum proteins, were measured on mass spectrometry. Albumin synthesis, expressed as a fractional rate, was 7.9% f 0.3%/day in the controls and 7.9% & l.l%/day in the cirrhotic patients. Albumin synthesis, expressed as an absolute rate, was lower in the cirrhotic group (cirrhotic, 119 5 17 mg/kg/day; controls, 146 5 8 mg/kg/day), but because of the relatively small number of patients the difference was not significant. However, the absolute rate of albumin synthesis significantly correlated with the Child-Turcotte score (p = 0.024) and its Pugh modification (p = 0.027). The rate of albumin synthesis also correlated with serum phenylalanine concentration but not with serum albumin concentration and intravascular albumin mass or with other clinical indexes of liver function or integrity when taken separately. However, the significant correlation between albumin synthesis and Child score suggests that albumin synthesis might be useful for the clinical judgment of patients with cirrhosis. (HEPATOLOGY 1993;18:292-297.) Albumin is the single most abundant serum protein; it is synthesized exclusively by the liver. Although albumin concentration was for many years considered a