Intrinsic hepatic clearance (K) of indocyanine green (ICG) is used as a quantitative measure of liver function. ICG is tightly bound to plasma proteins. The purpose of this study was to examine the effect of changes of plasma protein concentration on K in anaesthetized pigs with intact hepatic circulation. In addition, an attempt was made to evaluate the corresponding changes of the unbound intrinsic clearance of ICG. The plasma protein concentration was changed by exchange of plasma with either dextran-70 or donor pig plasma. Plasma albumin concentration was measured in a peripheral artery and changes of the concentrations of other plasma proteins were assumed to parallel those of albumin. ICG was given as a constant infusion and K was calculated from peripheral artery and hepatic vein concentrations of ICG according to the sinusoidal perfusion model. One experimental series comprised 3 measurement periods: From Period 1 to Period 2 (eight animals) albumin concentration was decreased by 36.6 +/- 6.5% (Mean +/- SD). This was associated with an increase of K of 32.8 +/- 28.8% (P = 0.004). From Period 2 to 3 (five animals) albumin was increased by 13.2 +/- 3.2% and K decreased by 18.5 +/- 8.3% (P = 0.03). In the second experimental series (eight animals), albumin concentration was increased by 21.6 +/- 10.3% and K decreased by 20.3 +/- 8.1% (P = 0.001). For both series, changes in albumin concentration were associated with oppositely directed changes of K in 20 out of 21 comparisons (P less than 0.001). Thus K depends not only on hepatocyte function but also on plasma protein concentration. This finding should affect interpretation of K when used as a liver function test. Changes of the unbound intrinsic clearance of ICG were examined indirectly by means of the K.a product (a: albumin concentration). According to the overall evaluation of the data the unbound intrinsic clearance of ICG was not affected by the changes in plasma protein concentration, but the results were internally inconsistent, apparently due to a time-dependency of the K.a product. We suggest this to be due to a slow but steady decrease of the 'background' K. After correction for the average decrease of K of 0.102% per min our data were in accordance with the hypothesis that the unbound clearance of K was enhanced by the binding protein(s) of ICG.