OBJECTIVES: To (a) describe differences in the outcome of cross sectional and longitudinal analysis on musculoskeletal complaints relative to age and work demands, and (b) to assess the entrance and drop out selection on musculoskeletal complaints among groups of employees relative to age and work demands. METHODS: A study population was selected on the basis of questionnaire data from periodical occupational health surveys of almost 45,000 employees collected between 1982 and 1993. From all companies within this data base that participated twice in company wide surveys four years apart, male employees were selected, and stratified for age and work demands. There were several populations: follow up (participation in both surveys); drop out (participation only in the first survey); entrance (participation only at the second survey); and two cross sectional populations (all participants at each survey). Prevalences of back complaints and turnover rates were analysed. RESULTS: Reported back complaints in the cross sectional analysis declined over the oldest age groups in heavy physical work versus a small increase in the longitudinal analysis. The age group 50-9 and back complaints were identified as predictors at the first survey for not participating at the second survey. Neither age nor work demands at the first survey indicated drop out among those employees with back complaints at the first survey. The effects of entrance selection on estimated prevalences were small. CONCLUSIONS: The results indicate that musculoskeletal disorders lead to selection out of work, affecting the validity of both cross sectional and longitudinal epidemiological studies. In future studies analyses of turnover figures on musculoskeletal complaints relative to work demands and age are recommended.