Botulinum neurotoxin is conventionally divided into seven serotypes, designated A–G, and is produced as large protein complexes through associations with non‐toxic components, such as hemagglutinin (HA) and non‐toxic non‐HA. These non‐toxic proteins dramatically enhance the oral toxicity of the toxin complex. HA is considered to have a role in toxin transport through the intestinal epithelium by carbohydrate binding and epithelial barrier‐disrupting activity. Type A and B HAs disrupt E‐cadherin‐mediated cell adhesion, and, in turn, the intercellular epithelial barrier. Type C HA (HA/C) disrupts the barrier function by affecting cell morphology and viability, the mechanism of which remains unknown. In this study, we identified GM3 as the target molecule of HA/C. We found that sialic acid binding of HA is essential for the activity. It was abolished when cells were pre‐treated with an inhibitor of ganglioside synthesis. Consistent with this, HA/C bound to a‐series gangliosides in a glycan array. In parallel, we isolated clones resistant to HA/C activity from a susceptible mouse fibroblast strain. These cells lacked expression of ST‐I, the enzyme that transfers sialic acid to lactosylceramide to yield GM3. These clones became sensitive to HA/C activity when GM3 was expressed by transfection with the ST‐I gene. The sensitivity of fibroblasts to HA/C was reduced by expressing ganglioside synthesis genes whose products utilize GM3 as a substrate and consequently generate other a‐series gangliosides, suggesting a GM3‐specific mechanism. Our results demonstrate that HA/C affects cells in a GM3‐dependent manner.