Inoculation of rabbit anti-idiotypic (anti-id) antibodies suppresses the subsequent appearance of a cross-reactive idiotype (CRI) associated with the anti-p-azophenylarsonate (anti-Ar) antibodies of A/J mice. Such suppressed mice produce normal concentrations of anti-Ar antibodies which lack the CRI, but against which anti-id antisera can be prepared. The anti-Ar antibodies of an individual, suppressed mouse do not in general share idiotype with anti-Ar antibodies of other A/J mice, either suppressed or nonsuppressed. The present experiments were undertaken to quantitate several "private idiotypes" in a large number of hyperimmunized A/J mice. Anti-Ar antibodies of three mice, suppressed for the CRI, were labeled with 125I and subjected to isoelectric focusing. Four single peaks, that were over 90% reactive with autologous antiid, were randomly selected for use as ligands in a radioimmunoassay, and ascitic fluids containing anti-Ar antibodies from 181 A/J mice were tested as inhibitors. Two of the four idiotypes could not be detected in any mouse other than the donor. The concentration of the idiotype was less than 1 part in 1,250 to less than 1 part in 25,000 of the anti-Ar antibody population; these are minimum values. A third idiotype was detected in 3 of the 181 mice, but at very low concentrations. The fourth idiotype was present in 28% of the mice, again at a low concentration. The data support the existence of a very large repertoire of anti-Ar antibodies in the A/J strain and are consistent with a process of random somatic mutation for generating diversity in hypervariable regions. It is proposed that the cross-reactive idiotype may be controlled by a germ line gene or a gene related to a germ line gene through a small number of somatic mutations; and that the idiotypes that were not detectable in other mice were the products of genes that had undergone extensive mutations, with a low probability of recurrence in other mice.