We describe the recurrence of a cross-reactive idiotope (CRIAD8) in antibody responses to different epitopes, and explore factors leading to its dominance in some responses, but not others. Serological and genomic DNA analyses showed that CRIAD8 is a marker of the CRIa heavy chain variable region (VH) that encodes the majority of anti-p- azobenzenearsonate (anti-ABA) antibodies. The independence of CRIAD8 from any particular light chain or antigen specificity was underscored by the fact that we could isolate hybridomas that secrete antitrinitrophenyl (TNP) antibodies expressing CRIAD8, with lambda 1 light chains. CRIAD8 is dominant in anti-ABA responses, recurrent but nondominant in anti-TNP and anti-chicken gammaglobulin responses, and is virtually absent in other antihapten responses, including that to p- azobenzenephosphonate (ABP), which contains an ABA-cross-reactive component (approximately 5-40%). Dominance in the anti-TNP response could not be achieved by immunization with doubly haptenated TNP-ABA- KLH (keyhole limpet hemocyanin), where the anti-ABA response was dominated by CRIAD8. These observations suggest that, while the CRIAD8 VH region is necessary for idiotypic dominance, it is not sufficient. Apparently, an additional specificity is required. Since immunization with ABA calls up anti-ABP antibodies that express CRIAD8, but not vice versa, it is possible that the additional specificity is ABA itself. This possibility imposes a new constraint on the specificity of the putative idiotype-specific regulation that may establish dominance in the CRIa system.