The present study performed a segregation analysis of a cohort of first-degree relatives (FDR) of glioma patients. The families with two or more gliomas were also expanded to determine if any more gliomas could be detected, and if any other types of cancers were associated. These glioma-prone families (n = 24/432) were extended to include first-, second- and third-degree relatives (n = 807) and a cohort was assembled, the standardized incidence risk for other types of cancer calculated and the pedigrees investigated for a possible mode of inheritance. A segregation analysis of the 2141 FDR in 297 families, performed using the Pointer software, did not clearly reject a multifactorial model chi(2)(3) = 6.13, P< 0.2. However, when letting all parameters be free, the recessive model provided the best fit. In the extended families, no increased risk of other types of cancer was found. This population-based study proposes that familial glioma occurs in about 5% of all glioma cases and that 1% have a possible autosomal dominant inheritance. This first segregation analysis performed in familial glioma must be cautiously interpreted, but an autosomal recessive gene provided the best fit, which could possibly explain 2% of all glioma cases.