OBJECTIVE To estimate the allele frequency of C9orf72 (G4C2) repeats in amyotrophic lateral sclerosis (ALS), frontotemporal lobar degeneration (FTLD), Alzheimer disease (AD), and Parkinson disease (PD). DESIGN The number of repeats was estimated by a 2-step genotyping strategy. For expansion carriers, we sequenced the repeat flanking regions and obtained APOE genotypes and MAPT H1/H2 haplotypes. SETTING Hospitals specializing in neurodegenerative disorders. SUBJECTS We analyzed 520 patients with FTLD, 389 patients with ALS, 424 patients with AD, 289 patients with PD, 602 controls, 18 families, and 29 patients with PD with the LRRK2 G2019S mutation. MAIN OUTCOME MEASURE The expansion frequency. RESULTS Based on a prior cutoff (>30 repeats), the expansion was detected in 9.3% of patients with ALS, 5.2% of patients with FTLD, and 0.7% of patients with PD but not in controls or patients with AD. It was significantly associated with family history of ALS or FTLD and age at onset of FTLD. Phenotype variation (ALS vs FTLD) was not associated with MAPT, APOE, or variability in the repeat flanking regions. Two patients with PD were carriers of 39 and 32 repeats with questionable pathological significance, since the 39-repeat allele does not segregate with PD. No expansion or intermediate alleles (20-29 repeats) were found among the G2019S carriers and AD cases with TAR DNA-binding protein 43-positive inclusions. Surprisingly, the frequency of the 10-repeat allele was marginally increased in all 4 neurodegenerative diseases compared with controls, indicating the presence of an unknown risk variation in the C9orf72 locus. CONCLUSIONS The C9orf72 expansion is a common cause of ALS and FTLD, but not of AD or PD. Our study raises concern about a reliable cutoff for the pathological repeat number, which is important in the utility of genetic screening.