Little is known about the comparability of postal and telephone survey modes in epidemiology. A cross-sectional, population-based study (n = 25,000) of lung disease was performed in 1998-1999 in two regions of Norway. Initial surveying was done by postal questionnaire. A 1% random sample (n = 171) of previous postal responders were resurveyed by telephone or cellular contact. The response rate was 67% on the telephone/cellular interview. Fewer incomplete answers were given by telephone than by mail. A lower prevalence was found by telephone for morning cough and exposure to passive smoking at work or home. Reproducibility was high for asthma, hay fever, wheezing, and attacks of breathlessness. Moderate reproducibility was seen for symptoms of chronic bronchitis. Reproducibility was low for indoor and work environment, although it was high for early life factors. Concordance coefficients were high for all continuous measures such as height, body weight, and pack-years. The authors conclude that the comparability between the postal and the telephone survey modes was good. The telephone survey mode gave more complete information. Survey mode may have a moderate effect on study results, depending on the specific questions asked.