The authors investigated the effects of glioma cells and pharmacological agents on the permeability of an in vitro blood-brain barrier (BBB) to determine the following: 1) whether malignant glia increase endothelial cell permeability; 2) how glucocorticoids affect endothelial cell permeability in the presence and absence of malignant glia; and 3) whether inhibiting phospholipase A2, the enzyme that releases arachidonic acid from membrane phospholipids, would reduce any malignant glioma-induced increase in endothelial cell permeability. Primary cultures of rat brain capillary endothelium were grown on porous membranes; below the membrane, C6, 9L rat glioma. T98G human glioblastoma, or no cells (control) were cocultured. Dexamethasone (0.1 microM), bromophenacyl bromide (1.0 microM), a phospholipase A2 inhibitor, or nothing was added to culture media 72 hours prior to assaying the rat brain capillary endothelium permeability. Permeability was measured as the flux of radiolabeled sucrose across the rat brain capillary endothelium monolayer and then calculated as an effective permeability coefficient (Pe). When neither dexamethasone nor bromophenacyl bromide was present, C6 cells reduced the Pe significantly (p < 0.05), whereas 9L and T98G cells increased Pe significantly (p < 0.05) relative to rat brain capillary endothelium only (control). Dexamethasone reduced Pe significantly for all cell preparations (p < 0.05). The 9L and T98G cell preparations coincubated with dexamethasone had the lowest Pe of all cell preparations. The Pe was not affected in any cell preparation by coincubation with bromophenacyl bromide (p > 0.45). These in vitro BBB experiments showed that: 1) malignant glia, such as 9L and T98G cells, increase Pe whereas C6 cells probably provide an astrocytic influence by reducing Pe; 2) dexamethasone provided significant BBB "tightening" effects both in the presence and absence of glioma cells; 3) the in vivo BBB is actively made more permeable by malignant glia and not simply because of a lack of astrocytic induction; 4) tumor or endothelial phospholipase A2 activity is probably not responsible for glioma-induced increased in BBB permeability; and 5) this model is useful for testing potential agents for BBB protection and for studying the pathophysiology of tumor-induced BBB disruption.