Background Public health monitoring depends on valid health and disability estimates in the population 65+ years. This is hampered by high non- participation rates in this age group. There is limited insight into size and direction of potential baseline selection bias. Methods We analyzed baseline non-participation in a register-based random sample of 1481 inner-city residents 65+ years, invited to a health examination survey according to demographics available for the entire sample, self-report information as available and reasons for non-participation. One year after recruitment, non- responders were revisited to assess their reasons. Results Five groups defined by participation status were differentiated: participants (N = 299), persons who had died or moved (N = 173), those who declined participation, but answered a short questionnaire (N = 384), those who declined participation and the short questionnaire (N = 324), and non-responders (N = 301). The results confirm substantial baseline selection bias with significant underrepresentation of persons 85+ years, persons in residential care or from disadvantaged neighborhoods, with lower education, foreign citizenship, or lower health-related quality of life. Finally, reasons for non-participation could be identified for 78 % of all non-participants, including 183 non- responders. Conclusion A diversity in health problems and barriers to participation exists among non-participants. Innovative study designs are needed for public health monitoring in aging populations.
