By means of a gel filtration on Sephadex columns the plasma corticoids can be separated into two fractions: namely a fraction bound to a macromolecular agent and an unbound fraction. When plasma is overloaded with cortisol before gel filtration, the binding capacity of the macromolecular agent can be evaluated. The bound corticoids can be measured quantitatively either flourimetrically or colorimetri-cally. The macromolecular binding agent behaves like transcortin in a series of experiments. Various exogenous and endogenous steroids are known to influence transcortin activity in vivo. Methylestrenolone and lynestrenol administered orally to normal males or to women with various gynecological conditions did not significantly increase the cortisol binding capacity. Intravenous adminstration of Metopirone lowered the cortisol binding capacity in normal subjects but not in a patient with hypophysial insufficiency. Attempts were made to remove steroids from plasma without destroying transcortin, in order to rule out any possible interference with the cortisol binding process by other steroids. Extraction of plasma with organic solvents either at room temperature or at minus 25[degree] C irreversibly destroyed most of the transcortin. Dialysis of plasma against a 1% solution of bovine albumin (40 volumes) for 12 h at 40[degree] C decreased the unconjugated corticoid level by 63%. This procedure did not alter transcortin activity in the plasma of normal adults, in umbilical cord blood, in pregnancy plasma or in the plasma of obese patients with a low cortisol binding capacity. However, the lowering in C.B.C. observed in the plasma of patients treated with Metopirone could be partly abolished by this procedure. Although the extraction experiments at low temperature suggested the existence of a lipid-link in the cortisol binding system of human plasma, this could not be demonstrated.