We investigated the in vivo pathogenic potential of murine autoimmunity to peptides of the glutamate/AMPA receptor subunit 3 (GluR3). Antibodies to GluR3 are found in human epilepsy, Rasmussen's encephalitis (RE). In our accompanying paper in this issue we found that murine antibodies to the GluR3B peptide (amino acids 372–395) bind neurons in culture, evoke GluR channel activity, and kill neurons in a complement-independent excitotoxic manner, mimicking the pathophysiologic effects of excess of glutamate. In the present study, we immunized four mouse strains (BALB/c, C3H/HeJ, SJL/J and C57BL/6) with the GluR3B peptide, and investigated the development of (1) anti-GluR3B antibodies; (2) anti-GluR3 T cells; (3) clinical symptoms and abnormal behaviour; (4) brain pathology. We found that BALB/c, C3H/HeJ and SJL/J mice strains developed high titres of anti-GluR3B antibodies. The low levels anti-GluR3B antibodies raised in C57BL/6 mice suggest that the genetic background of mice influences their ability to mount a humoral autoimmune response towards the GluR3B peptide. The GluR3B-immunized mice also developed anti-GluR3B T cells, and their splenocytes showed significantly biased frequencies of particular (Vβ11, Vβ7 and Vβ8) TCR Vβ families. Surprisingly, GluR3B-immunized mice also raised high anti-ssDNA humoral immunoreactivity. GluR3B-immunized mice exhibited multiple brain pathology, partially resembling that observed in RE, and subclinical behavioral abnormalities, but no epilepsy, even upon facilitating the entry of the autoreactive antibodies into the brain, by weakening the blood–brain barrier. Taken together, these results suggest that autoimmunity to the GluR3B epitope may account for the neuronal death and brain pathology seen in neurodegenerative diseases like RE, but may not be sufficient to underly epilepsy, at least not in mice.