A hexanucleotide repeat expansion in C9ORF72 has been established as a common cause of frontotemporal dementia (FTD). However, the minimum repeat number necessary for disease pathogenesis is not known. The aims of our study were to determine the frequency of the C9ORF72 repeat expansion in two FTD patient collections (one Australian and one Spanish, combined n = 190), to examine C9ORF72 expansion allele length in a subset of FTD patients, and to examine C9ORF72 allele length in 'non-expansion' patients (those with <30 repeats). The C9ORF72 repeat expansion was detected in 5-17% of patients (21-41% of familial FTD patients). For one family, the expansion was present in the proband but absent in the mother, who was diagnosed with dementia at age 68. No association was found between C9ORF72 non-expanded allele length and age of onset and in the Spanish sample mean allele length was shorter in cases than in controls. Southern blotting analysis revealed that one of the nine 'expansion-positive' patients examined, who had neuropathologically confirmed frontotemporal lobar degeneration with TDP-43 pathology, harboured an 'intermediate' allele with a mean size of only ∼65 repeats. Our study indicates that the C9ORF72 repeat expansion accounts for a significant proportion of Australian and Spanish FTD cases. However, C9ORF72 allele length does not influence the age at onset of 'non-expansion' FTD patients in the series examined. Expansion of the C9ORF72 allele to as little as ∼65 repeats may be sufficient to cause disease.