Twin registers have been established worldwide to study the roles of genes and the environment in health and behaviour. While questionnaire surveys are thought to be the most cost-effective way of collecting large amounts of data, low response rates can result in response bias. Many different strategies have been proposed to maximise response rates. A register of all multiple births occurring in Western Australia (WA) from 1980 onwards has been established using probabilistic record linkage techniques. Families who had not experienced the death of one or more of their multiples were invited to participate in the Western Australian Twin Child Health (WATCH) study, which studied the genetic and environmental determinants of childhood asthma and atopy. Several questionnaire designs and follow-up methods were assessed. We have shown that it was feasible to use a population-based register of multiple births to contact families for a questionnaire study. Questionnaire length, mode of follow-up, the number of responses required and the of participants all seemed to affect response.