Background: There is considerable uncertainty about work-relatedness and musculoskeletal disorders in general, and it is also not clear how physicians decide whether work has caused a disorder in an individual patient. Aims: To investigate physicians' beliefs about assessment of occupational attribution for work-related musculoskeletal disorders. Methods: A questionnaire survey was sent to 526 occupational physicians and 248 rheumatologists on: characteristics of cases seen, assessment of work attribution, definition of work-relatedness and threshold for case reporting. Continuous variables were analysed by mean, standard deviation, an independent two-sample t-test and the Mann-Whitney test. Mean and median values were calculated and Spearman's rank test was applied to ranked data. Results: Questionnaires were completed by 68% occupational physicians and 64% rheumatologists. Both groups of physicians believed that 'history of onset in relation to workplace changes' and 'symptoms consistent with work exposure' were the most important factors suggesting work attribution. They considered that the most important objective of a reporting scheme was detection of trends in disease incidence and that the most suitable criteria for defining work-relatedness was the probability that exposure at work 'more likely than not' caused the condition (mean 0.73; SD=0.17), in a perceived likelihood scale (0-1). Conclusion: There was a strong agreement between occupational physicians and rheumatologists on questions about work-relatedness and musculoskeletal disorders. The level of probability for concluding work-relatedness has been quantified. © The Author 2005. Published by Oxford University Press on behalf of the Society of Occupational Medicine. All rights reserved.