Stockholm Health of the Population Study is a cross-sectional study carried out from 1984-85. Postal questionnaires, telephone interviews and health interviews were used to get information from a sample of 5,199 persons, 18-64 years of age, on health status, risk exposures, healthcare consumption and social factors. Non-participation with respect to the postal questionnaire was 36.8%. With subsequent telephone interviews and an invitation to a health interview, non-participation was reduced to 17.8%. The estimated prevalence of daily smoking increased from 36.1% to 38.7. The non-responders had a higher prevalence of daily smoking in all sub-groups. This effect of the efforts to reduce non-participation differed socially. The prevalence of smoking for men, 40-64 years of age, who were reached by telephone was 60.3%. Male professionals and intermediate non-manual workers, 40-64 years of age reached by telephone had a prevalence of smoking, which was twice as high as for the responders of the questionnaire (62.5 and 26.8%, respectively). In the younger age-group, non-responders had the same socioeconomic pattern in smoking as the responders. Independent of socioeconomic group, there was a tendency of ill or disabled smokers to respond more quickly than healthy smokers. Using a postal questionnaire with a high non-response rate might lead to an overestimation of socioeconomic differences and an underestimation of smoking prevalence.