In the past, Huntington's disease was diagnosed when movement disturbances were present in concordance with a positive family history. Early motor signs, however, are often not found in a standard neurologic examination, and their expression is variable. Not until later stages of the disease could one be sure about the diagnosis. Since genetic diagnosis became available, the onset of symptoms and signs could be studied in the earliest phase. The aim of the study was to evaluate observer agreement of the assessment of early motor signs in Huntington's disease. A total of 17 gene carriers, 22 noncarriers, and eight partners were recorded on video performing several facial movements after instruction and engaging in a 5-minute conversation. Three experienced neurologists, unaware of the genetic status of the participants, judged and classified the video recordings independently. The cases with disagreement were judged a second time after the neurologists discussed these cases. The observer agreement after the first judgment is poor (kappas 0.09, 0.24, 0.45) and after the second judgment satisfactory (kappas 0.79, 0.90, 0.78). Consequences for clinical practice and research programs into early symptoms and signs are discussed.