Primary human hepatocytes were used to study bile salt hepatotoxicity and the hepatoprotective potential of ursodeoxycholate in vitro. Hepatocytes were obtained by collagenase perfusion of healthy human liver tissue and were treated with glycochenodeoxycholate for 24 hr 1 day after plating. Clear signs of cytotoxicity were observed at concentrations of about 100 μmol/L glycochenodeoxycholate. Toxicity was determined by release of alkaline phosphatase, γ-glutamyl transferase, AST, ALT or lactate dehydrogenase into the culture medium, by measuring DNA synthesis of the cultured liver cells and by testing the viability of the hepatocytes using trypan-blue dye exclusion. Addition of ursodeoxycholate, which by itself proved to be of little toxicity, significantly reduced the hepatotoxic effects of glycochenodeoxycholate: 72% ± 6% of the cells survived treatment with 500 μmol/L glycocheno-deoxycholate alone, but addition of 100 μmol/L ursodeoxycholate increased the survival rate to 87% ± 4% (p < 0.05). Moreover, all enzymes tested were secreted at a significantly lower level when ursodeoxycholate was present. Similarly, the cellular DNA synthesis was maintained at significantly higher levels as a result of ursodeoxycholate treatment. We conclude that (a) primary human hepatocytes are a suitable model for studying hepatotoxicity of bile salts in vitro, (b) ursodeoxycholate reduces hepatotoxicity of other bile salts and (c) ursodeoxycholate can act hepatoprotectively by itself (i.e., alteration of the metabolism of other bile salts is not necessarily required). (HEPATOLOGY 1990;12:486–491).