Two different cross-reactive idiotype (CRI) groups are distinguishable in the Ab response of A/J mice to the p-azobenzenearsonate (ABA) hapten: CRIA and CRIm. These two groups showed distinct patterns of relative dominance in the ensuing response depending on whether the inducing Ag was a T cell-dependent (TD) form of ABA, such as ABA-KLH or ABA-CGG, or a T-independent type 1 (TI-1) form, such as ABA-Brucella abortus or ABA-lipopolysaccharide (LPS), and on whether the response was elicited in vivo or in vitro. The CRI+ component of primary in vivo plaque-forming cell (PFC) responses to TD ABA Ags was largely (greater than 90%) CRIA+ as was, to a slightly lesser extent (greater than 75%) the CRI+ portion of secondary or hyperimmune serum Ab or PFC responses to the same Ags. In contrast, in vivo primary and hyperimmune PFC responses to ABA-Bru or ABA-LPS showed a significantly lower CRIA/CRI ratio, averaging 0.5-0.6, with some individual mice giving figures as low as 0.2, indicating predominance of CRIm over CRIA. Serological analysis of hyperimmune anti-ABA Abs from a group of 5 A/J mice immunized with ABA-Bru gave a figure of less than 0.5 for the CRIA/CRI ratio. The most striking disparity from the TD pattern was seen in primary in vitro PFC responses to the TI ABA Ags; here ratios of less than 0.2 were generally seen. Since T cell removal did not alter the Id pattern in the TI responses, CRIA-specific Ts cells do not account for the weak expression of CRIA in such responses. We propose a model that explains these results on the basis of differential expression of IdX dominance by two distinct B cell subpopulations--equatable to the Lyb- 5+ and Lyb-5- B cell subsets--along with differential relative activation of these subsets in different types of responses. Examination of anti-ABA PFC responses of F1 progeny of CBA/N and A/J mice to ABA-Bru lends support to this hypothesis since CRIA expression was significantly lower in mice with the xid defect.