This review seeks to explain the most exciting recent data concerning the nature of self/non-self discrimination by the immune system in a manner accessible to a biochemical readership. The nature of recognition in the two great lymphocyte families, B cells and T cells, is described with special emphasis on the nature of the ligands recognized by each. The history of the field of immunologic tolerance is surveyed, as are the key experiments on conventional mice which provided a conceptual framework. This suggested that tolerance was essentially due to 'holes' in the recognition repertoires of both the T and B cell populations so that lymphocytes competent to react to self antigens were not part of the immunologic dictionary. There were essentially two ways to achieve this situation. On the one hand, self antigens might 'catch' developing lymphocytes early in their ontogeny and delete the cell, a process of clonal abortion. On the other hand, self antigens might signal lymphocytes (particularly immature cells) in a negative manner, reducing or abolishing their capacity for later responses, without causing death. This process is referred to as clonal anergy. Evidence for both processes exists. Special emphasis is placed on a wave of experimentation beginning in 1988 which imaginatively uses transgenic mouse technology to study tolerance. Transgenic manipulations can produce mice which synthesize foreign antigens in a constitutive and/or inducible manner, sometimes only in specific locations; mice which possess T or B lymphocytes almost all expressing a given receptor of known specificity; and mice which are an immunologic time bomb in that the antigen is present and so too are lymphocytes all endowed with receptors for that antigen. These experiments have vindicated the possibility of both clonal abortion and clonal anergy in both T and B cell populations, the choice of which phenomenon occurs depending on a number of operational circumstances. For T cell tolerance, clonal abortion occurs if the self antigenic determinant concerned is present within the thymus; if not, clonal anergy is more likely. For B cell tolerance, the strength of the negative signal and therefore the choice between abortion and anergy depends on the molar concentration of the self antigen, the capacity for multivalent presentation to a B cell, and the affinity of the B cell's receptor for the antigen in question. Some B cells with low affinity for self antigens certainly escape censorship and remain capable of secreting low affinity anti-self antibodies, which however do no harm.(ABSTRACT TRUNCATED AT 400 WORDS)