Graves' disease may prove to be due to a genetic defect in immune surveillance or control, which then permits a randomly mutating thyroid-directed clone of T-lymphocytes to survive, interact with previously normal thyroid cell membrane antigen, and set up a localized cell-mediated immune response. The T-lymphocytes so interacting then cooperate with and direct groups of B-lymphocytes, which consequently produce TSI's: the stimulation of the thyroid parenchymal cells is a direct result of TSI stimulation. PHA stimulates Graves' lymphocytes to produce TSI's by first stimulating T-lymphocytes, which in turn interact with B-lymphocytes, which produce the TSI's. Normal thyroid antigen can stimulate sensitized lymphocytes to produce TSI's, which in turn can interact with normal thyroid cell membranes. This and other evidence suggests that there need not be any thyroidal antigenic change necessary to initiate hyperthyroidism. Stress may be related to the immunologic induction of the disease; it appears that remissions, other than those due to thyroid cell destruction, are immunologic. Exophthalmos may also prove to be an autoimmune disorder resulting from a closely related, but possibly not identical, defect in immunologic surveillance. Other related autoimmune diseases may result from similar related defects in immunologic surveillance, as well as from the random mutation of the appropriate forbidden clone.