Fragile X-associated tremor/ataxia syndrome, a neurodegenerative disorder associated with premutation alleles (55-200 CGG repeats) of the FMR1 gene, affects many carriers in late-life. Patients with fragile X-associated tremor/ataxia syndrome typically have cerebellar ataxia, intranuclear inclusions in neurons and astrocytes, as well as cognitive impairment. Dementia can also be present with cognitive deficits that are as severe as in Alzheimer's disease, however frontosubcortical type impairment is more pronounced in fragile X-associated tremor/ataxia syndrome. We sought to characterize the P600 and N400 word repetition effects in patients with fragile X-associated tremor/ataxia syndrome, using an event-related potential word repetition paradigm with demonstrated sensitivity to very early Alzheimer's disease. We hypothesized that the fragile X-associated tremor/ataxia syndrome-affected participants with poor declarative verbal memory would have pronounced abnormalities in the P600 repetition effect. In the event-related potential experiment, subjects performed a category decision task whilst an electroencephalogram was recorded. Auditory category statements were each followed by an associated visual target word (50% 'congruous' category exemplars, 50% 'incongruous' nouns). Two-thirds of the stimuli (category statement-target word pairs) were repeated, either at short-lag (approximately 10-40 s) or long-lag (approximately 100-140 s). The N400 and P600 amplitude data were submitted to split-plot analyses of variance. These analyses of variance showed a highly significant reduction of the N400 repetition effect (F = 22.5, P < 0.001), but not of the P600 repetition effect, in mild fragile X-associated tremor/ataxia syndrome (n = 32, mean age = 68.7, mean Mini-Mental State Examination score = 26.8). Patients with fragile X-associated tremor/ataxia syndrome had significantly smaller late positive amplitude (550-800 ms post-stimulus onset) to congruous words (P = 0.04 for group effect). Reduced P600 repetition effect amplitude was associated with poorer recall within fragile X-associated tremor/ataxia syndrome patients (r = 0.66) and across all subjects (r = 0.52). Larger P600 amplitude to new congruous words also correlated significantly with higher free recall scores (r = 0.37, P < 0.01) across all subjects. We found a correlation between the amplitude of late positivity and CGG repeat length in those with fragile X-associated tremor/ataxia syndrome (r = 0.47, P = 0.006). Higher levels of FMR1 mRNA were associated with smaller N400s to incongruous words and larger positive amplitudes (between 300 and 500 ms) to congruous words. In conclusion, event-related potential word repetition effects appear sensitive to the cognitive dysfunction present in patients with mild fragile X-associated tremor/ataxia syndrome. Their more severe reduction in N400 repetition effect, than P600, is in contrast to the reverse pattern reported in amnestic mild cognitive impairment and incipient Alzheimer's disease (Olichney et al., 2008).