Shiga-toxigenic Escherichia coli (STEC) causes severe gastrointestinal infections in humans that may lead to life-threatening systemic sequelae, such as the hemolytic uremic syndrome (HUS). Rapid diagnosis of STEC infection early in the course of disease opens a window of opportunity for therapeutic intervention, for example, by administration of agents that neutralize Shiga toxin (Stx) in the gut lumen. We previously developed a recombinant bacterium that expresses a mimic of the Stx receptor globotriaosyl ceramide (Gb3) on its surface through modification of the lipopolysaccharide (A. W. Paton, R. Morona, and J. C. Paton, Nat Med 6:265–270, 2000, http://dx.doi.org/10.1038/73111). This construct was highly efficacious in vivo, protecting mice from otherwise fatal STEC disease, but the fact that it is a genetically modified organism (GMO) has been a barrier to clinical development. In the present study, we have overcome this issue by development of Gb3 receptor mimic bacterial ghosts (BGs) that are not classified as GMOs. Gb3-BGs neutralized Stx1 and Stx2 in vitro with high efficiency, whereas alternative Gb3-expressing non-GMO subbacterial particles (minicells and outer membrane blebs) were ineffective. Gb3-BGs were highly efficacious in a murine model of STEC disease. All mice (10/10) treated with Gb3-BGs survived challenge with a highly virulent O113:H21 STEC strain and showed no pathological signs of renal injury. In contrast, 6/10 mice treated with control BGs succumbed to STEC challenge, and survivors exhibited significant weight loss, neutrophilia, and histopathological evidence of renal damage. Thus, Gb3-BGs offer a non-GMO approach to treatment of STEC infection in humans, particularly in an outbreak setting.