Background Lolland-Falster is the area of Denmark with highest mortality. However, clinical measurements from a population-based health survey showed prevalence of disease indicators similar to those in the rest of Denmark. The study aimed to disentangle this paradox.Methods The Lolland-Falster Health Study (LOFUS) took place in 2016textendash2020. We followed the 53 000 invited persons up for mortality from invitation date to 1 February 2023. Log-binomial regression was used to calculate relative risk (RR) of non-participation versus participation in LOFUS by subgroups of sex, age, municipality, residency group, household composition and economic status. Using Poisson regression, mortality rate ratio (MRR) was calculated between subgroups and within subgroups for non-participants versus participants for all-cause mortality and mortality from cancer, cardiovascular diseases, respiratory diseases, external causes and other diseases.Results One-third (36 of persons invited to LOFUS participated. Only modest differences were seen in participation across subgroups; the largest being an RR of 1.37 (95.35 to 1.40) for publicly supported versus self-supported persons. However, non-participants had higher mortality than participants (MRR 3.08, 95.82 to 3.37). This pattern was consistent across all subgroups and was found for both all-cause and cause-specific mortality.Conclusion The paradox we observed could partly be attributed to participation differences between subgroups. However, for the lack of population representativeness of the survey data, our study indicated within-group selection, measured by excess mortality of non-participants, to be much more important than between-group selection. One should therefore be cautious in using even weighted health survey data for prioritising health interventions.Data may be obtained from a third party and are not publicly available. Data are available upon request from Lolland-Falster Health Study, Statistics Denmark and the Danish Health Data Authority following Danish legislation.Data is available on request from Lolland-Falster Health Study, Statistics Denmark, and the Danish Health Data Authority following Danish legislation. Data are available upon request from Lolland-Falster Health Study, Statistics Denmark and the Danish Health Data Authority following Danish legislation.