Previous results from the literature pertaining to the molecular state of tetanus toxin are reported. By use of disc electrophoresis, gel filtration and ultracentrifugation, it is shown in this paper that tetanus toxin is likely to be constituted of subunits. Furthermore, it is postulated, starting from various lines of evidence, that the toxin should be recovered from the purification schedule as a dimer. The constituting monomeric units of the dimer are held together most probably by strong non-covalent interactions. Each monomer is proposed to consist again of two subunits of different sizes linked by one S-S bridge. The results which further this view are discussed together with the questions arising from this model.