I n these studies, we have used several approaches to systematically explore the contribution of transcellular vesicular transport (transcytosis) to the blood-to-bile movement of inert fluid-phase markers of widely varying molecular weight. First, under steady-state conditions, the perfused rat liver secreted even large markers in appreciable amounts. The bile-to-plasma (B/P) ratio of these different markers, including microperoxidase (B/P ratio = 0.06; mol wt = 1,879), insulin (B/P ratio = 0.09, mol wt = 5,000), horseradish peroxidase (B/P ratio = 0.04, mol wt = 40,000), and dextran (B/P ratio = 0.09, mol wt = 70,000), exhibited no clear ordering based on size alone, and when dextrans of two different sizes (40,000 and 70,000 mol wt) were studied simultaneously, the relative amounts of the two dextran species in bile were the same as in perfusate. Taurocholate administration produced a 71% increase in bile flow but little or no (0-20%) increase in the output of horseradish peroxidase, microperoxidase, inulin, and dextran. Second, under nonsteady-state conditions in which the appearance in or disappearance from bile of selected markers was studied after their abrupt addition to or removal from perfusate, erythritol reached a B/P ratio of 1 within 2 min. Microperoxidase and dextran appeared in bile only after a lag period of approximately 12 min and then slowly approached maximal values, whereas sucrose exhibited kinetically intermediate behavior. A similar pattern was observed after removal of greater than 95% of the marker from the perfusate. Erythritol rapidly reapproached a B/P ratio of 1, whereas the B/P ratio for sucrose, dextran, and microperoxidase fell much more slowly and exceeded 1 for a full 30 min after perfusate washout. Finally, electron microscopy and fluorescence microscopy of cultured hepatocytes demonstrated the presence of horseradish peroxidase and fluorescein-dextran, respectively, in intracellular vesicles, and fractionation of perfused liver homogenates revealed that at least 35-50% of sucrose, inulin, and dextran was associated with subcellular organelles. Collectively, these observations are most compatible with a transcytosis pathway that contributes minimally to the secretion of erythritol, but accounts for a substantial fraction of sucrose secretion and virtually all (greater than 95%) of the blood-to-bile transport of microperoxidase and larger markers. These findings have important implications with respect to current concepts of canalicular bile formation as well as with respect to the conventional use of solutes such as sucrose as markers of canalicular or paracellular pathway permeability.