The repeatability and validity of a questionnaire for upper limb and neck complaints were assessed in a population of 105 hospital outpatients with a range of upper limb and neck disorders (including cervical spondylosis, adhesive capsulitis, lateral epicondylitis, carpal tunnel syndrome and Raynaud's phenomenon). Subjects were asked to complete a modified Nordic-style upper limb and neck discomfort questionnaire on two occasions closely spaced in time. The repeatability of their responses was assessed by calculating a kappa coefficient (kappa), and the sensitivity and specificity of component items in the questionnaire were determined for specific diagnostic categories of upper limb and neck disorder. Symptom reports for pain in the upper limb and neck, pain interfering with physical activities, neurological symptoms and blanching were all found to be highly repeatable (kappa = 0.63-0.90). A number of regional pain reports proved to be very sensitive in relation to specific upper limb disorders, but, with the exception of reported finger blanching in patients with Raynaud's phenomenon, none proved to have a good specificity (range = 0.33-0.38). We conclude that a modified Nordic-style questionnaire is repeatable and sensitive, and is likely to have a high utility in screening and surveillance. However a complementary examination schedule of adequate specificity and repeatability is essential to establish a clinical diagnosis.