The possible relationship between Guillain-Barré syndrome (GBS) and cancer is still controversial and the existence of a paraneoplastic GBS remains unconfirmed. To better define whether there is a relationship between GBS and malignancy, we compared the observed and the expected number of patients with tumours in a population-based cohort of subjects with GBS. Clinical differences between GBS patients with or without malignancies were analysed. Data were obtained from the Piemonte and Valle d'Aosta Register for GBS (PARGBS) (years 1990-1998). GBS was diagnosed according to NINCDS criteria. The number of expected cases of malignancy in the PARGBS population was calculated using the incidence rate of all types of cancer (ICD codes 140-208) in Piemonte [1985-1987], and in the most important town of this region, that is Turin (years 1993-1997). In the nine-year period, 435 incident patients with GBS were found. Nine of them developed cancer in the six months preceding or following GBS; in seven of them, the diagnosis of cancer and GBS was concomitant. The expected number of malignant tumours was 3.7 (using the incidence in Piemonte) and 3.8 (using the incidence in Turin); therefore, the odds ratios were 2.43 (95 % CI, 1.11-4.62) and 2.37 (95% CI, 1.09-4.50), respectively (p<0.01). Although the cases with malignancies were clinically similar to the other cases of GBS observed through the Register, the mortality in GBS patients with cancer was higher and was the final cause of death in two patients affected by severe cancer. These results suggest a possible correlation between some cases of GBS and cancer. However, GBS in cancer patients does not meet all the criteria for paraneoplastic diseases.