No vaccine to protect against an estimated 238,000 shigellosis deaths per year is widely available. S. sonnei is the most prevalent Shigella, and multiple serotypes of S. flexneri, which change regionally and globally, also cause significant disease. The leading Shigella vaccine strategies are based on the delivery of serotype specific O-antigens. A strategy to minimize the complexity of a broadly-protective Shigella vaccine is to combine components from S. sonnei with S. flexneri serotypes that induce antibodies with maximum cross-reactivity between different serotypes. We used the GMMA-technology to immunize animal models and generate antisera against 14 S. flexneri subtypes from 8 different serotypes that were tested for binding to and bactericidal activity against a panel of 11 S. flexneri bacteria lines. Some immunogens induced broadly cross-reactive antibodies that interacted with most of the S. flexneri in the panel, while others induced antibodies with narrower specificity. Most cross-reactivity could not be assigned to modifications of the O-antigen, by glucose, acetate or phosphoethanolamine, common to several of the S. flexneri serotypes. This allowed us to revisit the current dogma of cross-reactivity among S. flexneri serotypes suggesting that a broadly protective vaccine is feasible with limited number of appropriately selected components. Thus, we rationally designed a 4-component vaccine selecting GMMA from S. sonnei and S. flexneri 1b, 2a and 3a. The resulting formulation was broadly cross-reactive in mice and rabbits, inducing antibodies that killed all S. flexneri serotypes tested. This study provides the framework for a broadly-protective Shigella vaccine which needs to be verified in human trials.