Here, we describe a structure-based approach to reduce the size of an antigen protein for a subunit vaccine. Our method consists of (i) determining the three-dimensional structure of an antigen, (ii) identifying protective epitopes, (iii) generation of an antigen fragment that contains the protective epitope, and (iv) rational design to compensate for destabilization caused by truncation. Using this approach we have successfully developed a second-generation Lyme disease vaccine. Outer surface protein A (OspA) from the Lyme disease spirochete Borrelia burgdorferi elicits protective immunity that blocks transmission of Borrelia from the tick vector to the vaccinated animal, and thus has been a focus of vaccine development. OspA has two globular domains that are connected via a unique single-layer beta-sheet. All anti-OspA monoclonal antibodies that block Borrelia transmission bind to conformational epitopes in the C-terminal domain of OspA, suggesting the possibility of using the C-terminal domain alone as a recombinant protein-based vaccine. The removal of ineffective parts from the OspA antigen may reduce side effects and lead to a safer vaccine. We prepared a C-terminal fragment of OspA by removing approximately 45% of residues from the N terminus. Although the fragment retained the native conformation and affinity to a protective antibody, its vaccine efficacy and conformational stability were significantly reduced with respect to full-length OspA. We successfully stabilized the fragment by replacing amino acid residues involved in buried salt-bridges with residues promoting hydrophobic interactions. The mutations promoted the vaccine efficacy of the redesigned fragment to a level comparable to that of the full-length protein, demonstrating the importance of the antigen stability for OspA's vaccine efficacy. Our strategy should be useful for further refining OspA-based vaccines and developing recombinant vaccines for other diseases.