The creatinine-method to estimate muscle mass is frequently used in clinical studies, although the validity of this approach is uncertain in patients with cirrhosis. In this study 102 patients with cirrhosis differing in cause, clinical state, liver, and renal function were investigated to determine whether reduced liver or renal function may explain in part the low levels of urinary creatinine excretion frequently observed in these patients. Muscle mass assessed by 24-hour urinary creatinine excretion was compared with anthropometrically obtained muscle mass calculated from arm muscle area (AMA), and with body cell mass (BCM) estimated by bioelectrical impedance analysis and total body potassium counting. In cirrhosis, the 24-hour urinary creatinine excretion was 10.4% and AMA was 19% lower than predicted values. The differences between the results obtained by different methods did not show any relation to parameters of liver function (ICG-t1/2, caffeine-t1/2, MEGX-test, cholinesterase) or the severity of liver disease (i.e., Child-Pugh score). In contrast, renal function was strongly correlated with the differences between creatinine- and anthropometric-muscle mass (r = .64, P < .001). At the same time, patients with normal renal function (62% of the whole population) had significantly higher creatinine (29.1 +/- 8.5 vs. 15.8 +/- 6 kg, P < .001) and anthropometric-muscle mass (22.4 +/- 6 vs. 17.9 +/- 5.3 kg; P < .01) than patients with reduced renal function (38% of the patients). In addition, significantly higher differences between measured and predicted values of urinary creatinine excretion (-0.389 +/- 0.33 vs. 0.06 +/- 0.31 g/24 h; P < .001) and of AMA (13.2 +/- 12 vs. 7.2 +/- 12 cm2; P < .03) were found in the subgroup with impaired renal function. In conclusion, renal dysfunction but not reduced liver function systematically affects the urinary creatinine method for the estimation of skeletal muscle mass in cirrhosis.