The study was based on a questionnaire and included 209 white-collar workers and 241 blue-collar workers in eight companies. The questionnaire deals with musculoskeletal symptoms, psychosocial and physical load factors. The relationship between job factors and neck, shoulder and low-back symptoms was analysed. The Nordic Musculoskeletal Questionnaire (NMQ) was supplemented with an additional question regarding whether or not the symptoms were believed to be related to work. The associations between the psychosocial factors and musculoskeletal symptoms were substantially higher when solely work-related symptoms were included in the analyses; these associations were of the same magnitude as that between several physical work load factors and work-related musculoskeletal symptoms. This study suggests that calculations of associations based solely on the NMQ, without the inclusion of questions to elicit reports of symptoms that are believed to be work-related, could have a powerful effect-masking consequence.