The biased use of V beta 8.2 and V beta 6 in rats by encephalitogenic T cells specific for the S72-89 and S87-99 epitopes of guinea pig basic protein (Gp-BP) has allowed the use of anti-V beta antibodies and synthetic TCR peptides for treatment of experimental autoimmune encephalomyelitis (EAE). Striking V gene biases also occur in human autoimmune diseases, raising the question of to what degree these biases reflect potentially pathogenic T cells. To address this question, we evaluated the expression of the EAE-associated marker V beta 8.2 and V beta 6 molecules in the periphery, spinal cord (SC), and cerebrospinal fluid (CSF) during the course of EAE, in unselected, IL-2-expanded, and Gp-BP-restimulated populations. In CSF cells, there was a strong bias for the marker V beta before the onset of EAE, but this bias was not enhanced by IL-2, which skewed the CSF population to > 80% CD8+ T cells. In SC, the marker V beta were expressed optimally during the onset of EAE, even in unselected cells, and this bias could be enhanced sequentially by IL-2 expansion and Gp-BP restimulation. During the recovery phase, however, the marker V beta 8.2 bias was obfuscated by the appearance of a heterogeneous V beta T cell population. Biased expression of the marker V genes was not detected in unselected or IL-2-expanded peripheral cells at any time during EAE. These data suggest that peripheral T cells bearing the disease-relevant V genes first appeared in CSF before disease onset and then migrated to SC beginning on the first day of clinical signs. During the recovery phase of the disease, these cells were diluted by an influx of T cells bearing other V beta genes, requiring restimulation with Gp-BP to observe the V beta 8.2 bias. These data have important implications for the interpretation of V beta gene biases that have been reported in human autoimmune diseases.