Abstract. Corticosteroid binding globulin (CBG) was detected by a specific radioimmunoassay in mixed saliva (25.4 ± 4.0 μg/l, mean ± sem) and in pure, uncontaminated parotid fluids (17.4 ± 2.7 μg/l) at resting flowrates of approximately 500 μl/min and 50 μl/gland per min, respectively. In parotid fluids collected at stimulated flow-rates of between 300–1000 μl/gland per min, CBG could not be detected. This observation suggests the direct flow-rate-dependent transfer/secretion of CBG in saliva. When cortisol was measured (RIA) in dilution experiments in both mixed saliva and parotid fluids using phosphate buffer at pH 7.4 as diluent, a protein-binding effect analogous to that found in plasma samples was observed. However, this effect was abolished if a known CBG inhibitor, phosphate:citrate buffer at pH 4, was used as the diluent in the assay. A bound fraction of cortisol was found in both mixed saliva (14.0 ± 4.0%) and parotid fluid samples (12.3 ± 1.3%) by equilibrium dialysis. These findings appear to contradict the currently accepted notion that specific plasma steroid binding proteins, and hence the protein-bound steroids, are absent in uncontaminated saliva; and that their presence in mixed saliva is the consequence solely of contamination by gingival fluid and/or plasma from mouth or gum abrasions. We conclude that both protein-bound and free steroids are present in uncontaminated saliva and that salivary total and plasma free steroid concentrations are not identical.