We have used a well-defined idiotypic system, the cross-reactive idiotype of A strain (CRIA) (Ab1) idiotype generated in A/J mice injected with arsonate coupled to keyhole limpet hemocyanin (ARS-KLH), to determine the frequency of precursors for auto-anti-idiotypic antibodies (auto-Ab2) in naive and immunized A/J mice by limiting dilution analysis after polyclonal activation by lipopolysaccharide. In naive animals, the precursor frequencies of auto-Ab2 B cells were below the limit of sensitivity of the technique in the majority of A/J mice, and could be detected in only 20% of the animals. Upon immunization with ARS-KLH, a large increase in auto-Ab2 precursor frequency was observed. This shift in frequency was not found when A/J mice were injected with KLH alone, or when BALB/c mice, which do not express the CRIA idiotype, were injected with ARS-KLH. To study the functional role of the auto-Ab2 B cells, we injected neonatal A/J mice with polyclonal rabbit Ab3 antibodies directed against a recurrent idiotype of auto-Ab2. Thereafter, these mice were injected with ARS-KLH. Although the anti-arsonate response level was normal, the CRIA Ab1 expression was reduced tenfold. Thus, the suppression of auto-Ab2 affects Ab1 dominance. We further show that the presence of maternal Ab1 can strongly modify the immune response of the offspring by inducing higher levels of the idiotype after immunization. Furthermore, IgM anti-arsonate antibodies were detected before immunization with antigen. From these data, we conclude that the affinity of antigen alone cannot explain the dominance of CRIA. Network selection is important in the shaping of the available repertoire.