The action of bile salts upon the rat blood-brain barrier (BBB) was assessed in the absence of energy-yielding metabolism. Brains were perfused in situ with a Ringer solution for 5 min followed by a 1 min perfusion containing either sodium deoxycholate (DOC), taurochenodeoxycholate (TCDC), or Ringer/DNP. The integrity of the BBB was then determined by perfusing with the radiotracer [14C]mannitol for 2.5 min. Alternatively, the brains were perfusion fixed for ultrastructural assessment. At 0.2 mM DOC, the BBB remained intact and the cerebral ultrastructure was similar to the controls. At 1 mM and above, disruption of the BBB became evident. At 2 mM, the cerebral cortex became severely vacuolated, with damaged endothelium and collapsed capillaries. With TCDC, BBB disruption occurred at 0.2 mM without any apparent ultrastructural damage to the microvasculature. Following 2 mM TCDC, similar, but less widespread, structural changes to the 2 mM DOC-perfused animals was apparent. Opening of the BBB occurred at a concentration lower than that required to cause lysis of either red blood cells or cultured cerebral endothelial cells. It is proposed that the effect of bile salts at concentrations of 1.5 mM and above is largely due to their lytic action as strong detergents on endothelial cell membranes, but that at lower concentrations a more subtle modification of the BBB occurs.