The memory cells of two antibody-forming cell clones had receptors of the IgM class, even though the clones had been producing IgG1 or IgG2a anti-2,4-dinitrophenyl antibodies for 9-15 months previously (on exposure to antigen). Thus a phenotypic switch in heavy chain constant region evidently occurred after re-exposure of these memory cells to antigen. To show that, we first removed the clonal cells' surface immunoglobins by "capping" and "stripping", with class- or subclass-specific antisera. Then, to assay their remaining receptor activity, the cells were incubated with antigen in vitro, washed and transferred (together with carrier primed cells) to irradiated recipients, and their antibody responses to this in vitro boost were assayed by iselectric focusing. Pretreatment with anti-mu serum, as well as with anti-Fab(kappa), prevented the responses of the IgG1 and IgG2a clones to an in vitro boost, while anti-gamma1 and anti-gamma2a antisera had no effect. An antiserum to the putative mouse IgD also had no effect. The anti-mu serum failed to react with the IgG1 and IgG2A clonal serum antibodies in the test tube. Some other contaminating clones were suppressed completely only by the anti-Fab serum. This result strongly suggests that switching in class commitment may occur during the differentiation of memory cells to antibody producers, and may therefore be antigen-dependent. It also implies that some apparently naive cells with surface IgM may, in reality, be B memory cells.