Hexanucleotide repeat expansions within the C9orf72 gene are the most important genetic cause of amyotrophic lateral sclerosis (ALS) and frontotemporal dementia (FTD). The difficulty of developing a precise method to determine the expansion size has hampered the study of possible correlations between the hexanucleotide repeat number and clinical phenotype. Here we characterize, through a new non-radioactive Southern blot protocol, the expansion size range in a series of 38 ALS and 22 FTD heterozygous carriers of >30 copies of the repeat. Maximum, median and modal hexanucleotide repeat number were higher in ALS patients than in FTD patients (P< 0.05 in all comparisons). A higher median number of repeats correlated with a bigger range of repeat sizes (Spearman's ρ = 0.743, P = 1.05 × 10(-11)). We did not find any correlation between age of onset or disease duration with the repeat size in neither ALS nor FTD mutation carriers. Clinical presentation (bulbar or spinal) in ALS patients did not correlate either with the repeat length. We finally analyzed two families with affected and unaffected repeat expansion carriers, compared the size of the repeat expansion between two monozygotic (MZ) twins (one affected of ALS and the other unaffected), and examined the expansion size in two different tissues (cerebellum and peripheral blood) belonging to the same FTD patient. The results suggested that the length of the C9orf72 repeat varies between family members, including MZ twins, and among different tissues from the same individual.