New insights into the autoimmune basis for Addison's disease have come from identification of at least three P450 cytochrome enzymes as autoantigens, each having distinct associations with Addison's disease as part of the APS type 1 or 2 syndrome. Enzymes are tissue-restricted proteins which are the frequent targets of autoimmunity in other organ-specific diseases (Editorial, 1992), and it seems likely that further P450 enzymes could be involved in the pathogenesis of other components of these syndromes. How adrenal damage is initiated remains unclear. Adrenal autoantibodies may have a pathogenic role, as yet obscure, or could arise secondary to T cell-mediated tissue damage, although it seems highly likely that the same autoantigen provokes cell-mediated and humoral autoimmunity. Sharing of autoantigens between ovary and adrenal glands, particularly the side-chain cleavage enzyme, is one explanation for the close association of ovarian failure and Addison's disease, but other, more common forms of ovarian autoimmune disease exist. Their further definition will come from identification of the autoantigens involved. By analogy with animal models, T cell-mediated injury is likely to be central to pathogenesis. The evidence for antibodies blocking hormone receptors in premature ovarian failure is meagre at present, but the availability of recombinant LH and FSH receptors should clarify this issue. HLA-DR3 is associated with almost all autoimmune endocrinopathies, and this is particularly striking in APS type 2. However, there is no such association with APS type 1; the most likely genetic candidate in this condition is at a locus controlling T cell development. Although the adrenal and ovarian autoimmune processes in APS type 1 and 2 may be distinct, the characterization of the gene involved in APS type 1 will have major implications for our understanding of autoimmune endocrine disease.