Objectives: To quantify the relative contributions of behavioural, environmental and psychological factors to the vocal health of teachers, and to describe the relationships using structural equation modelling, with a view to identifying preventive action. Method: A cross-sectional survey of teachers across 69 primary and secondary schools was conducted. In total, 217 responses were analysed. Teachers self-reported on: the quality of their voice; the frequency with which they perform a series of voice-related behaviours; the quality of the environment in which they work; the feelings they have about their vocal health; and an anxiety rating measured by the State-Trait Anxiety Inventory (STAI) questionnaire. A structural equation modelling approach was used to estimate the associated effects. Results: The hypothesized model described the data well. It emerged that voice-related behaviours, the teachers' work environment and the presence of trait anxiety had a significant influence on vocal health. Moreover, the model demonstrated that the quality of the voice is related strongly to how respondents feet about the condition of their voice, which, in turn, had an indirect reciprocal effect on the quality of teachers' vocal health. Conclusion: The model demonstrates the important contributions of psychological and behavioural variables to vocal health. Of the six independent variables that impact directly on `vocal dysfunction', three were found to be statistically significant. These were the voice-related `behaviours' that teachers perform, the environment in which teachers work, and `trait anxiety present' as measured by the STAI. The implications of the results are considered in relation to rethinking policy and practice with the intention of identifying preventative actions to improve the vocal health of professional educators. (C) 2007 The Royal Institute of Public Health. Published by Elsevier Ltd. All rights reserved.