Anchor polymerase chain reaction has been applied to the study of human T cell receptor beta-chain repertoire in peripheral blood. The use of this technique has demonstrated that considerable variation in V(beta) and J(beta) usage exists, both within and between individuals. Particular V(beta) families, including V(beta)6, V(beta)4 and V(beta)12 are commonly utilized, while families such as V(beta)10, V(beta)11 and V(beta)15 are rare in all individuals studied. Marked interindividual variation in V(beta) usage was detected for V(beta)12 and V(beta)4. Biased usage of J(beta) elements is a prominent feature of peripheral repertoire, while there is no evidence for preferential V(beta)-J(beta) recombination events. Biased J(beta) usage in expressed V(beta)-D(beta)-J(beta)-C(beta) transcripts, subject to selection, was the same as that in aberrant, unselected D(beta)-J(beta)-C(beta) transcripts, implying that bias resulted from events relating to rearrangement itself, in the absence of selection. N-region diversity showed some evidence for preferential insertion of deoxyguanosine, consistent with the action of terminal deoxytransferase. No P-nucleotide incorporation was seen in association with intact J(beta) elements. These data provide evidence of some of the variation in human T cell receptor beta-chain repertoire and provide a basis for comparisons with sequences which may be obtained in autoimmune and superantigen-mediated diseases.