The association of relative weight with various indices of musculoskeletal morbidity was studied in 902 Finnish white-collar and blue-collar workers. Data were collected in the years 1973 and 1983 through a postal questionnaire, by medical examination, and clinical assessment of the condition of the musculoskeletal system. In the overweight group (relative weight greater than or equal to 120 percent), higher mean levels in all musculoskeletal scores (the Musculoskeletal Symptoms Score, the Upper/Lower Extremities Score, the Low Back Score and the Musculoskeletal Findings Score) were observed, compared with the normal weight group. There was no evidence, however of rheumatic complaints in weight-bearing joints being more common in overweight individuals, contrary to the 'wear and tear' hypothesis. In a 10-year follow-up the changes in the musculoskeletal scores were analysed by stepwise multiple regression. Baseline relative weight explained 6.3 percent of the variation in the score based on clinical assessment. Neither baseline weight nor weight change were significant predictors of change in the other scores. No association was observed between overweight and lumbosacral disorders either in the cross-sectional study or in the 10-year follow-up.