As the available diagnostic criteria (National Institute of Neurological and Communicative Disorders and Stroke, NINCDS) for Guillain-Barré syndrome (GBS) do not permit inclusion of clinical variants (CV) of GBS, there are few data on their occurrence and few reports of the overall incidence of the disease. A population-based study in the local health district of Ferrara, Italy in 1981-1993 selected cases fulfilling both NINCDS criteria (NINCDS GBS cases) and CV. The incidence of CV was 0.35 per 100,000 person-years (95% CI: 0.15-0.68), 0.32 when age-adjusted to the Italian population. No difference was found between CV and NINCDS GBS for male/female ratio, mean age at onset, elevated CSF protein content, seasonal pattern, or mean time delay from first neurological symptom to maximal severity. A higher frequency of antecedent infections for CV and more frequent serious disease at the nadir time for NINCDS GBS were found. A complete recovery was more frequent for CV than NINCDS GBS, but no difference was found regarding good outcome (defined by a satisfactory recovery and resumption of normal functional life). Since most findings were similar for NINCDS GBS and CV cases, they may have similar underlying pathological mechanisms. When diagnostic criteria for GBS include CV, the overall disease incidence in the Ferrara district increases from 1.87 to 2.21 cases per 100,000 person-years (the contribution of CV to the overall incidence of GBS is 15.7%). The currently available diagnostic criteria for GBS, although useful for field studies, may be too restrictive as they can entail the loss of about 15% of cases.