Lewis rats were used to determine the encephalitogenic activity of myelin basic protein of different species and of 45-residue fragments of basic protein. Basic protein from guinea pigs was more active than that from rats, and the fragments from the two species showed the same order of activity. Bovine basic protein was the least active of the intact proteins, and the respective fragment was inactive. Studies of serum-binding capacity did not support the hypothesis that blocking antibody played a role in this biological variation, whereas consideration of the amino acid sequences of the three fragments suggested that differences in primary structure, operating either at the sensitization or the effector phase of the immune response, could account for the variation.