Huntington's disease (HD) produces progressive and ultimately widespread impairment of brain function. Neostriatal atrophy alone cannot account for whole-brain losses seen postmortem, and the mutant huntingtin protein and its neuropathologic sequelae are evident throughout the brain. Whole-brain atrophy quantification encompasses the totality of mutant huntingtin's effects on brain volume and may be useful in tracking progression in trials. We studied whole-brain atrophy in HD using a 2-year follow-up design, with three annual MRI scans. We recruited 20 control subjects, 21 premanifest mutation carriers, and 40 patients with early HD and used the brain boundary shift integral to study rate and acceleration of atrophy. Among subjects with an acceptable quality 2-year scan pair, age- and gender-standardized mean brain atrophy rate was greater (P < 0.001) in the patients with HD (n = 21; 0.88%/yr; 95% confidence interval: 0.62-1.13%/yr) than that in controls (n = 13; 0.16%/yr; 0.00-0.32%/yr). In the 12 patients with early HD in whom acceleration could be directly assessed there was evidence (P= 0.048) of acceleration year-on-year (mean acceleration = 0.69% yr(-2); 95% confidence interval: 0.01% yr(-2) to 1.37% yr(-2)), although this was not formally significantly different from that in controls (n = 7, P = 0.055). Statistically significantly increased atrophy rates and acceleration were not seen overall in the premanifest group, who were on average 18 years from predicted disease onset. We conclude that the study of whole-brain atrophy has the potential to inform our understanding of the neurobiology of HD and warrants further study as one means of assessing the outcomes of future clinical trials.