Among other outcomes, volunteer surveys are useful for evaluating conservation success and determining priorities for management actions. However, biases that can originate from untargeted and weakly structured surveys can undermine the utility of the data gathered. Identifying and rectifying biases and problems with such data require an understanding of the behaviour of volunteers. We explored the characteristics of volunteer behaviour using bird surveys conducted in south-western Australia, and evaluated how volunteer behaviour affects the quantity and quality of data when volunteers are unconstrained in their selection of survey sites. We related the home-range and site-fidelity of 172 volunteers to the availability of habitat and bird species. Habitat selection by volunteers was assessed using avian species-accumulation curves, which identified 12 habitats for which avian species inventories were <95% complete. Volunteer biases resulted in skewed representation of birds in the resulting dataset. We recommend the minimum sampling effort required for reliable species-richness estimates in each habitat, and suggest ways to achieve spatial representativeness by using different behavioural types. Volunteers with high site-fidelity (often locals) produce high species detection rates, and are useful for long-term monitoring or surveying in less-favoured habitats close to urban areas. Roaming volunteers (often tourists) with large home-ranges are useful for threatened species surveying and can fill gaps far from urban areas, but might require incentives to visit unfavoured habitats, given their high habitat and bird selectivity. By studying volunteer behaviour, we can set realistic goals to achieve a comprehensive dataset useful for research, management and conservation planning.