Inhibition of canalicular bile acid efflux by medications is associated with clinical liver toxicity, sometimes in the absence of major liver effects in experimental species. To predict the hepatotoxic potential of compounds in vitro and in vivo, we investigated the effect of clinical cholestatic agents on [3H]taurocholic acid transport in regular and collagen-sandwich cultured human hepatocytes. Hepatocytes established a well-developed canalicular network with bile acid accumulating in the canalicular lumen within 15 min of addition to cells. Removing Ca2+ and Mg2+ from the incubation buffer destroyed canalicular junctions, resulting in bile acid efflux into the incubation buffer. Canalicular transport was calculated based on the difference between the amount of bile acid effluxed into the Ca/Mg2+-free and regular buffers with linear efflux up to 10 min. Hepatocytes cultured in the nonsandwich configuration also transported taurocholic acid, but at 50% the rate in sandwiched cultures. Cyclosporin A, bosentan, CI-1034, glyburide, erythromycin estolate, and troleandomycin inhibited efflux in a concentration-dependent manner. In contrast, new generation macrolide antibiotics with lower incidence of clinical hepatotoxicity were much less potent inhibitors of efflux. An in vivo study was conducted whereby glyburide or CI-1034, administered iv to male rats, produced a 2.4-fold increase in rat total serum bile acids. A synergistic 6.8-fold increase in serum total bile acids was found when both drugs were delivered together. These results provide methods to evaluate inhibitory effects of potentially cholestatic compounds on bile-acid transport, and to rank compounds according to their hepatotoxic potential.