OBJECTIVES: To assess whether healthcare professionals correctly incorporate the relevance of a favourable test outcome in a close relative when determining the level of risk for individuals at risk for Huntington's disease. DESIGN AND SETTING: Survey of clinical geneticists and genetic counsellors from 12 centres of clinical genetics (United Kingdom, 6; The Netherlands, 4; Italy, 1; Australia, 1) in May-June 2002. Participants were asked to assess risk of specific individuals in 10 pedigrees, three of which required use of Bayes' theorem. PARTICIPANTS: 71 clinical geneticists and 41 other healthcare professionals involved in genetic counselling. MAIN OUTCOME MEASURES: Proportion of respondents correctly assessing risk in the three target pedigrees; proportion of respondents who were confident of their estimate. RESULTS: 50%-64% of respondents (for the three targets separately) did not include the favourable test information and incorrectly estimated the risks as being about equal to the prior risks; 77%-91% of these respondents were "sure" or "completely sure" that their estimations were correct. Twenty of the 112 respondents correctly estimated the risks for all three target pedigrees. CONCLUSIONS: Clinical geneticists and genetic counsellors frequently use prior risks in situations where Bayes' theorem should be applied, leading to overestimations of the risk for an individual.