Liquid from canine airway surfaces was absorbed onto filter paper strips and analyzed. In resting conditions, tracheal surface liquid was hyperosmolal (330 mosmol/kg H2O) compared to plasma with raised Na(+1) (158 meq/l), Cl(-1) (134 meq/l), K(-1) (28 meq/l), and HCO3(-1) (32 meq/l) concentrations. The volume collected was increased by repetitive sampling, a response blocked by atropine, or by methacholine injection. Compared to nose breathing, tracheal surface liquid osmolality was increased by 10 min of mouth breathing (410 mosmol/kg H2O). Surface liquid from 0.5-cm diameter bronchi was nearly isosmolal (304 mosmol/kg H2O) with plasma in resting conditions, with Na and Cl concentrations lower that plasma (120 and 106 meq/l, respectively), and K(-1) (52 meq/l), and HCO3(-1) (50 meq/l) concentrations higher than those of plasma or tracheal liquid. Although the K(+1) in tracheal fluid approaches the value for electrochemical equilibrium, K(+1) in fluid from the bronchi and HCO3(-1) in both regions cannot be accounted for by passive forces. The regional difference in osmolality supports the concept that the higher osmolality of tracheal liquid reflects evaporative water loss from this site. The transepithelial osmotic gradient generated by evaporative water loss may be a driving force for hydration of the tracheal surface.