In such cases, there always is an inherent risk of false positive result due to chance or to bias such as occult stratification. Although this risk cannot be excluded for each separate association, the following points are noteworthy as, all taken together, they strongly argue in favor of the actual participation of the C3 gene in the genetic susceptibility to MTLE-FS+ and to FS. First of all, the very conservative Bonferroni method was applied to correct against statistical effects of multiple testing. Moreover, the false discovery rate analysis also pointed at the existence of true positives. Secondly, protective effect of HAP4 against MTLE-FS+, while significantly displayed in samples of small sizes, was replicated in independent patients and controls groups. That rare alleles can indeed influence susceptibility to common disorders is increasingly recognized; as an example, the 15q13 microdeletion variant that was present in 1% of patients with idiopathic generalized epilepsy and that was virtually absent from controls, is associated with strong epileptogenic effect. Thirdly, the protective effect of 11-containing HAP5 haplotype in a first series of FS patients, was consistent with the protective effect of 11 allele alone in a novel and independent FS population. Moreover, C3 maps right within the 19p critical area where linkage of autosomal dominant FS trait has been reported. Fourthly, there is an obvious clinical overlap between MTLE-FS+ and FS patients and the C3 susceptibility gene was found significantly associated in these independent populations of patients. Last but not least, within the C3 promoter region, influenced the promoter activity of C3 in vitro. Indeed and as is always desirable in genetic association studies, we showed here that the newly discovered GF100472 microsatellite polymorphism had functional consequences. Generally, CA-repeats are purine-pyrimidine alternating sequences and hence have the potential to assume Z-DNA conformation, a left-handed double-helix structure that is thermodynamically unfavorable compared to B-DNA conformation. Z-DNA conformation has been described as negatively affecting transcriptional activity. Consistently, we showed that GF100472 regulated C3 promoter activity in a lengthdependent manner: the longer the CA-repeat, the lower the transcriptional activity. GF100472 may at least partly influence genetic risks to epilepsy and epileptic seizures by modulating the relative amounts of C3 transcripts. Interestingly, the IL-1b cytokine that had shown genetic association with FS and with MTLE also influences C3 promoter activity through regulatory binding elements situated in the C3 promoter region. The promoter activities associated with the GF100472 alleles did not vary by more than 20%. Such a moderate effect is consistent with the participation of such alleles as susceptibility factors in multigenic disorder, as compared with the direct and more dramatic effects of WZ8040 mutations in monogenic disorders.