Although well-characterized in the lung, the role of platelet-activating factor (PAF) in inflammation in the central nervous system is undefined. Using rabbit models of meningitis and pneumonia, PAF was found to induce significant blood-brain barrier permeability and brain edema at doses five times lower than those required to generate leukocyte recruitment to the subarachnoid space. Both leukocytosis and increased vascular permeability occurred in response to PAF in the lung. Antibody to the CD-18 family of leukocyte adhesion molecules inhibited leukocyte recruitment in response to PAF in the brain (greater than 80%); a similar level of inhibition in the lung required treatment with a combination of a PAF receptor antagonist (L-659,989) and anti-CD18 antibody. Treatment with L-659,989 decreased abnormal cerebrospinal fluid cytochemical values induced by intracisternal challenge with pneumococci but not Haemophilus influenzae, indicating a special role for PAF in pneumococcal disease. Antibodies directed at phosphorylcholine, a unique, shared determinant of bioactivity of PAF and pneumococcal cell wall, obviated the inflammatory potential of both agents. However, no evidence for a direct PAF-like activity of pneumococcal cell wall components was detected in vitro by bioassay using platelets or neutrophils. It is concluded that PAF can induce inflammation in the subarachnoid space. In brain, PAF effects appear to be mediated through CD-18-dependent events, while in lung, PAF effects independent of CD-18 are also evident. At both sites, PAF is of particular clinical importance during inflammation induced by pneumococci apparently due to a unique proinflammatory relationship between the pneumococcal cell wall teichoic acid and PAF.