Variables potentially influencing the rate of return of postal questionnaires sent to 311 coalminers who had left their place of employment between 1977 and 1982 were investigated. Three measures of respiratory health (obstruction, restriction, and presence of chronic bronchitis symptoms), as well as the potentially confounding factors of age, education, marital status and smoking habits, were included in a logistic model used to analyze returned questionnaires. It was hypothesized that miners with impaired respiratory function or with respiratory symptoms would respond to a survey more readily than miners without impairment or symptoms, which would introduce a bias toward overestimation of ill health. In order to test this hypothesis, questionnaires were sent to miners for whom baseline respiratory parameters were already known. None of the measures of respiratory health, nor any of the other potentially confounding variables, except age, correlated with speed of questionnaire return. These results suggested that when age is correlated with health status, any age bias in sampling would involve a health status bias. It may be possible in many cases to adjust for age bias by standardization, especially in those cases where age bias is suspected.