Symptomatic patients with Huntington's disease may have reduced glucose metabolism in the caudate nuclei. We used positron emission tomography and [18F]fluorodeoxyglucose to study cerebral glucose metabolism in 95 subjects: 58 clinically asymptomatic (chorea-free) subjects at risk for Huntington's disease, 10 symptomatic patients with the disease, and 27 controls. All the symptomatic patients had marked reductions in caudate glucose metabolism. Despite a normal structural appearance on computed tomography, caudate glucose metabolism was bilaterally reduced in 31 percent of the subjects at risk (18 of 58). Using each at-risk subject's age and the sex of the affected parent, we averaged individual risk estimates for the development of Huntington's disease for this group and predicted that the probability of having the clinically unexpressed Huntington's disease gene should be 33.9 +/- 6.0 percent for the group. Thus, there was excellent agreement between the predicted percentage of carriers of the Huntington's disease gene (33.9 +/- 6.0 percent) and the percentage with metabolic abnormalities of the caudate nuclei (31 percent). These results indicate that the measurement of glucose metabolism may allow the observation of the pathophysiologic effects of the Huntington's disease gene during the natural development of the disease. It may also provide a direct means to monitor the response to experimental treatments during both the clinically asymptomatic and the symptomatic phases of the disorder.