We used double-label quantitative autoradiography to measure blood flow (with 131I-iodoantipyrine) and blood-to-tissue transport of 14C-alpha aminoisobutyric acid, AIB) in thirteen 9L gliosarcomas transplanted intracerebrally into Fischer-344 rats. Microscopically, the typical pattern of 9L tumor growth was observed: a solid central tumor mass surrounded by extensive perivascular invasion. The averaged mean whole tumor transfer constant, K, of AIB in the 9L tumors was 33 +/- 15 (+/- SD) microliters/g/min. The averaged mean value of blood flow, F, was 72.2 +/- 27.3 ml/100 g/min. In brain around tumor (BAT), K (13 +/- 4 microliters/g/min) was lower than in the solid tumor, but was still 6-9 times that of tumor-free brain. F in BAT (115.9 +/- 64.6 ml/100 g/min) was comparable to values in tumor-free cortex in the same hemisphere. Values of K and F were used to calculate a net extraction fraction (En) for different regions of brain and tumor. The value of En of AIB in normal cortex was 0.003, in BAT En was 0.02, and in whole tumor the value was 0.09. The delivery of water-soluble compounds in 9L brain tumors is limited by the permeability/surface area characteristics of the tumor capillaries. The properties of blood-to-tissue transport and blood flow of 11 different brain tumor models are compared, and discussed with regard to the choice of brain tumor models for drug delivery research. The 9L brain tumor model is comparable to other transplanted rat brain tumor models, although the extent of tumor cell invasion into BAT makes this model distinctive. However, with regard to blood-to-tissue transport the 9L model differs from autochthonous models and transplanted human glioma models. We discuss guidelines for selecting brain tumor models with which to study the problem of drug delivery to brain tumors.