We investigated whether chromosome 9 open reading frame 72 hexanucleotide repeat expansion (C9orf72 expansion) size in peripheral DNA was associated with clinical differences in frontotemporal degeneration (FTD) and amyotrophic lateral sclerosis (ALS) linked to C9orf72 repeat expansion mutations. A novel quantification workflow was developed to measure C9orf72 expansion size by Southern blot densitometry in a cross-sectional cohort of C9orf72 expansion carriers with FTD (n = 39), ALS (n = 33), both (n = 35), or who are unaffected (n = 21). Multivariate linear regressions were performed to assess whether C9orf72 expansion size from peripheral DNA was associated with clinical phenotype, age of disease onset, disease duration and age at death. Mode values of C9orf72 expansion size were significantly shorter in FTD compared to ALS (p = 0.0001) but were not associated with age at onset in either FTD or ALS. A multivariate regression model correcting for patient's age at DNA collection and disease phenotype revealed that C9orf72 expansion size is significantly associated with shorter disease duration (p = 0.0107) for individuals with FTD, but not with ALS. Despite considerable somatic instability of the C9orf72 expansion, semi-automated expansion size measurements demonstrated an inverse relationship between C9orf72 expansion size and disease duration in patients with FTD. Our finding suggests that C9orf72 repeat size may be a molecular disease modifier in FTD linked to hexanucleotide repeat expansion.