Normal New Zealand White rabbits were used to compare theoretical brain concentrations (based upon pharmacokinetic modeling) with actual experimental concentrations of BCNU following intraarterial (IA) or intravenous (IV) infusions. IA infusion therapy for brain tumor patients has been promising based upon theoretical predictions but of limited effectiveness clinically. Experimentally-measured rabbit carotid artery flow rates (63.9 +/- 3.4 ml/min) [mean +/- 1 sem] and BCNU systemic clearances (197 +/- 10.2 ml/min) predicted a theoretical IA advantage of 4.1 +/- 0.2. Ipsilateral brain concentrations of BCNU during and after IA infusions (20 mg/min/m2 over 15 minutes) were: 16.2 +/- 2.9, 19.0 +/- 3.9, 20.3 +/- 2.8, 4.8 +/- 2.5, 2.1 +/- 1.5, and 1.7 +/- 1.6 micrograms/gm brain at 5, 10, 15, 25, 35, and 45 minutes after infusion start. Mean concentrations at same time points in contralateral hemisphere (IA infusions) were: 7.1 +/- 1.8, 9.0 +/- 1.8, 10.3 +/- 0.7, 4.2 +/- 1.4, 2.2 +/- 1.2, 2.0 +/- 1.5 micrograms/gm brain. Concentrations in either hemisphere during IV infusions were similar to contralateral hemisphere during IA infusions. Comparison of ipsilateral: contralateral hemisphere ratios during and after IA infusions were: 3.2 +/- 0.4, 2.6 +/- 0.3, 2.2 +/- 0.3, 1.1 +/- 0.3, 1.0 +/- 0.4, and 0.9 +/- 0.3 at the same time points. Although these data show higher drug concentrations with IA infusions, actual values were considerably less than predicted by theoretical modeling. This discrepancy between theoretical and experimental results emphasizes need for further study of causes and remedies so that IA therapy can achieve better drug concentrations with less toxicity.