The best-characterized autoimmune T-cell response is that to myelin basic protein (MBP). MBP has classically been regarded as a sequestered antigen that does not cause negative selection. This view has been fostered by the observation that T-cell receptor-transgenic T cells that are specific for the "immunodominant determinant" on the molecule, MBP:Ac1-11, persist as naive cells in MBP-expressing H-2u mice. The same T cells, however, can cause autoimmune pathology once they have been primed by environmental stimulation to become memory cells. Once the autoimmune response to Ac1-11 has been engaged, determinant spreading occurs and second-wave T-cell responses that are specific for weaker, "cryptic" determinants like MBP:121-140 develop. Although the nature of these cryptic determinants has been enigmatic, recent studies using MBP-/- mice have provided new insights. These studies showed that MBP is not a sequestered antigen, but one that causes negative selection; as MBP:121-140 is actually the immunodominant determinant in MBP-/- mice, it tolerizes high avidity clones in MBP+/+ mice, making it appear cryptic. Based on this new information, we attempt here to redefine the MBP-specific repertoire within the theoretical framework of the threshold model for negative selection, and we propose a model of shifting T-cell activation thresholds to explain how ignorant/naive T cells can become effector cells of autoimmune pathology and why this effector cell repertoire spreads.