Pertussis toxin (PT) is a key protective antigen in vaccine- and natural immunity-mediated protection from Bordetella pertussis infection. Despite its importance, no PT-neutralizing epitopes have been characterized structurally. To define neutralizing epitopes and identify key structural elements to preserve during PT antigen design, we determined a 3.6 A cryo-electron microscopy structure of genetically detoxified PT (PTg) bound to hu11E6 and hu1B7, two potently neutralizing anti-PT antibodies with complementary mechanisms: disruption of toxin adhesion to cells and intracellular activities, respectively. Hu11E6 bound the paralogous S2 and S3 subunits of PTg via a conserved epitope, but surprisingly did not span the sialic acid binding site implicated in toxin adhesion. High-throughput glycan array analysis showed that hu11E6 specifically prevents PTg binding to sialylated N-glycans, while a T cell activation assay showed that hu11E6 blocks PTg mitogenic activities to define the neutralizing mechanism. Hu1B7 bound a quaternary epitope spanning the S1 and S5 subunits, although functional studies of hu1B7 variants suggested that S5 binding is not involved in its PT neutralization mechanism. These results are the first to structurally define neutralizing epitopes on PT, improving our molecular understanding of immune protection from B. pertussis and providing key information for the future development of PT immunogens.