The objective of this study was to estimate probability of survival of Huntington's disease (HD) patients in Serbia as a function of CAG repeat length and selected demographic variables. This follow-up study was carried out at the Institute of Neurology, Clinical Centre of Serbia, Belgrade, 1982-2004. The study group consisted of 112 HD patients. The significant inverse correlation was found between CAG repeat length and age at onset of HD (r = -0.732, P = 0.001) and age at death (r = -0.760, P = 0.001). The cumulative probabilities of survival in a five, ten, fifteen, and twenty-years' period were 90.9, 63.2, 10.3 and 4.5%, respectively. Higher survival probabilities were registered in female patients, as well as in those with older age at onset and lower number of CAG repeat length (</=46). The Cox regression analysis showed that significantly poorer outcome of HD in our population was related to younger age at onset (HR-hazard ratio = 1.9; P = 0.047), and larger CAG numbers (HR = 2.4; P = 0.071). The female sex was statistically significantly associated with longer survival (HR = 0.4; P = 0.007). These data might be of some importance for further exploration of natural history and prognosis of HD.