A self-administered questionnaire with questions related to the physical workload at video display units (VDUs) was tested for intra- and intermethod reliability among 36 men and 64 women working with research and documentation. The results showed an acceptable test-retest agreement. The answers to three questions were validated by direct measurements: the locations of keyboard and mouse on the table, and the distance between elbow and keyboard heights when working. The self-reported locations of keyboard and mouse showed good agreement with the direct measurements. A revised version of the questionnaire was used at a telecommunication laboratory to study work postures and musculoskeletal symptoms. Among CAD operators with identical work tasks, the women (n = 67) reported a higher prevalence of musculoskeletal symptoms than did the men (n = 475). Calculated prevalence rate ratios (women/men) for musculoskeletal symptoms ranged from 1.4 (low back) to 3.4 (left elbow). The CAD operators with at least 5.6 hours of mouse use/week (median) reported more symptoms in the arms than did the CAD operators with fewer hours' work. Operators with the mouse located outside an "optimal" area on the table reported more symptoms from shoulder joints (upper arms), shoulders (scapular), elbows, and wrists than did operators with the optimal mouse location. Thus, long hours of work with the mouse, as well as working with the mouse non-optimally located on the table, seemed to be risk factors for upper-limb symptoms.