ABSTRACT Human transcortin or a high affinity, low capacity corticosteroid-binding protein from the plasma of various species has been used for corticosteroid analysis. The relative specificity varies with the species so that selection of the appropriate plasma source is advantageous. Endogenous steroids which interfere in the assay may be limited by administration of a synthetic glucocorticoid prior to collection of the binding plasma or they may be removed by charcoal adsorption. Interference from albumin (and other non-specific binding proteins) may be minimized by dilution of the plasma. Further dilution may be attained if the transcortin concentration is raised by estrogen stimulation. Association of cortisol to human transcortin is virtually instantaneous and is limited in practice only by the rates of solution and diffusion. At 37° the rate of dissociation is also rapid (half-time ca. 10 sec.) so that only a short incubation time (5–15 min) is needed. With the almost universally accepted usage of non-equilibrium methods of separating bound from unbound steroid, it is imperative that separation be performed at low temperatures where the rate of dissociation is much less (half-time 25 min). There is insufficient information to decide which of the many variants of the technique is best for one particular corticosteroid. They all assume competing steroids to be negligible and, therefore, none (but one) of the variants incorporate a purification step (solvent partition or chromatography). In certain instances (late pregnancy, adrenogenital syndrome, metapyrone test) this can lead to varying degrees of error.