Blood-brain barrier disruption with a hyperosmolar agent, mannitol, has previously been demonstrated to increase intracerebral methotrexate levels in rats. To determine the optimum conditions for blood-brain barrier disruption without producing neurological sequelae, adult Sprague-Dawley rats were infused with mannitol via the internal carotid artery at rates varying from 0.25 to 0.5 ml.s-1.kg-1. Methotrexate and Evans blue were used as markers of blood-brain barrier disruption. The optimum rate of mannitol that produced blood-brain barrier disruption without neurological sequelae was 0.25 ml.s-1.kg-1 for 20 s. The duration of blood-brain barrier opening was maximal for approximately 5 min and then rapidly reversed. Methotrexate levels on the mannitol-infused side were four to five times that of the noninfused hemisphere. Light microscopy and electron microscopy did not demonstrate any consistent changes that could be attributed to blood-brain barrier disruption nor did it elucidate the mechanism. This model should prove useful in the investigation of the treatment of intracerebral tumors with blood-brain barrier disruption. This study shows that maximal intracerebral methotrexate levels were obtained when methotrexate was infused before or within 5 min of the mannitol infusion.