1. Plasma extravasation through the endothelium of blood vessels is an integral component of the inflammatory response and is dependent to a large extent on the inflammatory mediator bradykinin (BK). We studied plasma extravasation induced by BK perfusion (BK-induced PE) of the rat knee joint before and after various interventions that affect the sympathetic supply to the knee joint. We tested the hypothesis that plasma extravasation is dependent on the local sympathetic terminal supply to the synovia but not directly dependent on more proximal activity in the sympathetic neurons. As a control we used plasma extravasation induced by platelet activating factor (PAF), which acts directly on the endothelium of the blood vessels, that is, its action is independent of any innervation. Plasma extravasation into the knee joint cavity was determined spectrophotometrically by measuring, over time, the concentration of Evans blue dye extravasation into the joint perfusate following intravenous injection of the dye. 2. Surgical sympathectomy at the lumbar level (L2-L4), performed 4 and 14 days previously, reduced BK-induced PE by approximately 55-70%. 3. Decentralization of the lumbar sympathetic chain (cutting the preganglionic axons that innervate the postganglionic neurons to the hindlimb), interruption of the lumbar sympathetic chain during infusion of BK, or coperfusion of tetrodotoxin into the knee joint cavity did not reduce BK-induced PE. All three interventions abolish the activity in the sympathetic neurons but leave the peripheral postganglionic terminals in the joint capsule intact. 4. Surgical sympathectomy and decentralization did not affect plasma extravasation induced by the intra-articular perfusion with PAF. 5. Electrical stimulation of the lumbar sympathetic chain at frequencies of 0.25-5 Hz, which probably also significantly decreases blood flow through the joint capsule, reduced basal plasma extravasation, BK-induced PE and PAF-induced PE. This reduction was frequency dependent and was almost maximal at a stimulation frequency of 1 Hz. 6. In conclusion, BK-induced PE into the rat knee joint is dependent on the presence of intact sympathetic postganglionic nerve terminals innervating the joint capsule and not directly dependent on excitation of these neurons. However, electrical stimulation of the sympathetic neurons reduces the level of plasma extravasation, presumably because of vasoconstriction and decrease of blood flow through the joint capsule. These results indicate that peripheral action of inflammatory mediators on terminals of sympathetic neurons produces a facilitative effect on vascular permeability, whereas centrally generated excitation of these neurons, which depresses blood flow (vasoconstrictor function), decreases plasma extravasation. The effect on blood flow is presumed to occur at the precapillary resistance vessels by vesicular release of transmitter(s). The facilitative effect on permeability occurs at the venules and includes inflammatory-mediator-stimulated, non-vesicular-dependent production and release of a chemical substance (probably prostaglandin E2). Whether both functions are represented in the same class of sympathetic postganglionic neuron or in distinct ones remains to be elucidated.