Self-tolerance is generally induced by intrathymic clonal deletion of T cells with reactivity directed to antigens synthesized within the thymus (Kappler et al. 1987, Kisielow et al. 1988). It may also be induced in peripheral T cells when these encounter antigens unique to extra-thymic tissues. Two transgenic models have been particularly useful in the study of peripheral self tolerance: in one model, a known antigen is expressed in a particular extra-thymic site; in the other, the T-cell repertoire is predominantly reactive to this antigen. We, and others, have shown that expression of class I or II MHC molecules in defined extra-thymic sites leads to a state of T-cell tolerance. To account for this, we have proposed two hypotheses which have different implications for autoimmune disease. According to one, tolerance is imposed by deletion or functional silencing of specific high-affinity cytolytic T cells; alternatively, the target cell for tolerance induction may be a regulatory IL-2-producing T-cell, rather than the effector cell itself. To distinguish between these hypotheses it is essential to examine the fate of T cells which have the potential to react to the transgene product. Since the frequency of such T cells is low and there is no dominant clonotype for H-2Kb, which is the class I molecule we used, it was necessary to create double transgenic mice by mating class I transgenic mice with transgenic mice whose T-cell pool was compared of cells reactive to H-2Kb and could be detected by an antibody directed to the TCR. Initial studies showed that such T cells did persist despite the presence of antigen to which they may be reactive. If these double transgenic mice can be shown to be tolerant, they will offer a rich source of tolerant T cells for detailed investigation of their phenotype and fate, and they will be most useful in enabling us to probe the mechanisms responsible for the induction of peripheral self tolerance. Transgenic mouse technology has also been used successfully to unravel the genetic influences which may lead to or prevent autoimmunity. In particular, we have prevented autoimmune diabetes in the nonobese diabetic mouse by introducing a non-NOD MHC class II gene and further work is implicating the failure of intrathymic positive selection of a protective cell as one step in the pathogenesis of diabetes in NOD mice.