Occurrence of transcytosis in pleural mesothelium was verified by measuring removal of labeled macromolecules from pleural liquid in experiments without and with nocodazole. To this end, we injected 0.3 ml of Ringer-albumin with 750 microg of albumin-Texas red or with 600 microg of dextran 70-Texas red in the right pleural space of anesthetized rabbits, and after 3 h we measured pleural liquid volume, labeled macromolecule concentration, and, hence, labeled macromolecule quantity in the liquid of this space. Labeled albumin left was 318 +/- 28 microg in control and 419 +/- 17 microg in nocodazole experiments (means +/- SE); hence, whereas ventilation was similar its removal was greater (P < 0.01) in control experiments. Labeled dextran left was 283 +/- 10 microg in control and 381 +/- 21 microg in nocodazole experiments; hence, whereas ventilation was similar its removal was greater (P < 0.01) in control experiments. These findings indicate occurrence of transcytosis from the pleural space. Liquid removed by transcytosis was 0.05 ml/h. This amount times unlabeled albumin concentration under physiological conditions (10 mg/ml) times lumen-vesicle partition coefficient for albumin (0.78) provides fluid-phase albumin transcytosis: approximately 203 microg. h(-1) kg(-2/3). Transcytosis might contribute a relevant part of protein and liquid removal from the pleural space.