The splenic B-cell repertoire of unimmunized C57BL/6 mice can be examined for anti-(4-hydroxy-3-nitrophenyl)acetyl (NP) B cells of relatively high affinity by using a dual strategy. First, limiting numbers of splenocytes are polyclonally activated by Escherichia coli lipopolysaccharide and a mixture of interleukins 2, 4, and 5 in the presence of 3T3 filler cells, thus ensuring that many B-cell clones switch to IgG1 antibody production. Second, an enzyme-linked immunosorbent assay is geared to register only higher-affinity antibody by (i) detecting only bivalent IgG1 antibody and ignoring IgM and (ii) using a lowly substituted NP-conjugated protein as the capture layer. Naive spleens contain very few higher-affinity anti-NP B cells thus defined, but thymus (T)-dependent immunization causes the appearance of approximately 10(5) per spleen within 2 weeks. The development of these clonable anti-NP antibody-forming cell precursors can be virtually eliminated by a single injection of 1 mg of soluble, freshly deaggregated NP2-human serum albumin (HSA). This toleragen works not only if injected prior to challenge immunization, but even if given up to 6 days later. Soluble HSA works partially but not nearly as well as NP2-HSA, suggesting the possibility that the toleragen must act on T and B cells. NP conjugated to irrelevant carriers achieved partial tolerance in only one of four experiments. The studies demonstrate the need for continuing T-cell help throughout the process of memory B-cell generation. They also show that those recently activated T cells involved in this process can be silenced in vivo by soluble toleragen.