Valid generalizations of results from population-based epidemiological surveys requires knowledge about how representative the sample is. The Copenhagen Center for Prospective Population Studies have assessed mortality on the basis of pooled data from three research programmes in the region of Copenhagen. In two of the studies, subjects were randomly selected, using the Danish Central Population Registry, within certain age groups and area-restricted sectors of the Greater Copenhagen. In the third study, men employed in 14 companies participated. Participation rates were between 78% and 87% in the three programmes. Standardized mortality rates (SMR) were calculated in relation to mortality rates in the municipality of Copenhagen and in the whole country in three age groups and the two genders. SMR values in the whole sample including non-participants were similar to rates for Copenhagen in the Copenhagen City Heart Study, whereas mortality rates in the Glostrup Population Studies were similar to rates for the whole country. The mortality rates among participants were lower than in the whole sample, and differences existed in relation to region and selection criteria of the cohorts. The Copenhagen Male Study, where only employed men were included, showed the lowest mortality rates, and higher rates were found in the study from the central part of the City (the Copenhagen City Heart Study) compared to the study from the suburbs (the Glostrup Population Studies). The difference between mortality rates in the cohorts and in Copenhagen City decreased with increasing age. The SMR converged towards 1.00 with increasing observation time. In conclusion, high participation rates were found in all three studies, resulting in SMR values for participants only slightly lower than in the source population in the two randomly selected samples, but 30% lower values in the sample of employed men. As mortality rates in the total samples including non-participants were markedly higher than among the participants, generalizations of results for participants to the whole population should be made with caution, especially during the first years of observation.