The ventricular space of rabbits was perfused with a low-viscosity silicone oil for the purpose of (1) collecting freshly secreted cerebrospinal fluid (c.s.f.) uninfluenced by diffusional exchanges with the brain and (2) studying passage of solutes from the blood into the brain, uncomplicated by exchanges with c.s.f. The freshly secreted c.s.f. appeared as fine droplets suspended in the less dense silicone, and accumulated at the bottom of the collected silicone. Studies on the penetration of (24)Na from blood into this fluid indicated that considerable exchanges with the brain had occurred between its secretion and collection, in spite of this method of collection. The second objective was attained, in that the exchanges between the freshly secreted fluid and the brain were quantitatively insufficient to affect the measure of kinetics of uptake by brain from the blood. In consequence, it was possible to demonstrate unequivocally that the increased uptake by brain of (131)I, when treated with perchlorate, was due to inhibition of an active process occurring across the blood-brain barrier. Other studies, involving ventriculo-cisternal perfusion with artificial c.s.f., lent further support to this concept. (131)I distribution is some 32% of the brain weight, a figure close to the ;chloride-space'.