Experiments to test the relationship between the epitopes on a protein antigen recognized by T and B cells in their collaboration to produce antibody cannot rely solely on hapten‐carrier models. In the present work we used E. coli β‐galactosidase, a molecule whose tertiary and quaternary epitopes have been well characterized, as the model antigen. T helper cells were raised by stimulating mice with the intact or the denatured molecule or with any of several β‐galactosidase cyanogen bromide pep‐ tides. In a series of in vitro helper T cell assays we confronted the various T popula tions with B cells preimmunized with the native antigen, and we tested their capacity to help production of (a) binding antibodies and (b) antibodies directed to single conformational epitopes, characterized by their capacity to protect the enzyme from heat denaturation or to activate defective P‐galactosidase. According to our results, (a) equivalent T cell help can be provided by T helper cells primed with native or denatured antigen, even for the production of “conformational” antibodies; (b) one of the peptides (CB‐18) is most efficient in raising help for binding antibodies; and (c) two peptides (CB‐20 and CB‐21) rank highest in priming T helper cells for the even‐ tual production of protecting and activating antibodies, respectively. Thus, not every β‐galactosidase‐specific T helper cell is useful in providing help to B cells specific for any particular epitope on the molecule, but rather preferential pairings exist, possibly governed by a proximity rule.