A retrospective case study of 33 Guillain-Barré syndrome (GBS) patients with the antibody to the ganglioside N-acetylgalactosaminyl GD1a (GalNAc-GD1a) was made to investigate the clinical features of GBS with this antibody. Patients were classified into three groups: (i) 25 with IgG antibody (group G, titre >/= 1 : 40); (ii) 16 with high-titre IgG antibody (group G-high, titre >/=1:320; selected from group G patients), and (iii) eight with IgM antibody but without elevation of IgG (group M, normal range <1:40 for both IgM and IgG). The control group consisted of 72 GBS patients without anti-GalNAc-GD1a antibody. Compared with the control group, the G-high and G group patients were characterized as having had antecedent gastrointestinal infection (87% and 72% versus 31%, both P < 0.001), uncommon cranial nerve involvement (19% and 36% versus 54%, P = 0.02 and 0.2, respectively), distal-dominant weakness (94% and 68% versus 36%, P < 0.001 and P = 0.01, respectively) and no sensory signs (81% and 60% versus 25%, P < 0.001 and P = 0.003, respectively). Electrophysiological findings indicative of axonal dysfunction were significantly more common in the G-high and G group patients (63% and 52% versus 14%, both P < 0.001). The pure motor variant that showed neither sensory signs nor abnormalities in sensory conduction studies was also more frequent in these groups (44% and 32% versus 9%, both P < 0.001). IgG anti-GalNAc-GD1a antibody may be a marker of the pure motor and the axonal variants of GBS, and therefore it, as well as anti-GM1 antibody, must be investigated in these forms in order to diagnose and understand the variants. By contrast, mild weakness, frequent facial palsy (75%) and a high incidence of IgM anti-GM2 antibody reactivity (88%) were characteristic of group M, indicating that the GBS in that group resulted from a different immune mechanism from that in the G group.