Conformational constraints were imposed on a peptide epitope from Chlamydia trachomatis to improve its ability to elicit antibodies that cross-react with native antigen. Appropriate constraints were discovered by a strategy that required no prior knowledge of the epitope's native conformation. First, we constructed a library of 3.2 x 10(5) peptides in which the epitope's contact residues were subject to random conformational constraints, each constrained peptide being fused genetically to the surface of a filamentous phage vector. Next, we selected phage displaying the most native-like peptides in the library by affinity purification with antibodies that bind the epitope only in its native conformation. Finally, we immunized mice with the selected phage and titered the resulting antisera against both whole cells and unconstrained peptide. The ratio of anti-cell titer to anti-peptide titer, which reflects the channeling of the antibody response to the native epitope, was up to five times higher for affinity-selected phage than for unselected peptide phage. In this case, therefore, "antigenic fitness," the ability of a peptide to bind antibodies specific for native epitope, correlated with "immunogenic fitness," its ability to elicit antibodies that are effective against the native antigen on an invading pathogen. If the correlation is general, surveying thousands or millions of peptides for antigenic fitness with phage display technology may be a simple but effective pre-screen for immunogenic fitness, which is costly to assess directly.