While participating in a controlled study of baclofen as protective therapy, 39 Huntington's disease (HD) patients underwent measurements of horizontal saccade latency and velocity, repeated longitudinally over a 2-year period. Significant worsening of saccade latency and of mean velocity was detected in untreated patients. Although individual variation was great, initial velocity impairment was found to be more prominent in younger patients. Factors are identified that may affect the rate of decline in supranuclear oculomotor function, including age and the severity of illness at the time of initial assessment. We propose that serial quantitative measurement of saccade performance is a useful clinical marker of the rate of disease progression against which the efficacy of treatments may be tested.