A way of measuring patients' beliefs about the origin of their symptoms would allow the investigation of important questions concerning the consultation process and its outcome. The purpose of this study was to develop an instrument that could measure the beliefs about symptoms of patients attending their general practitioner and to demonstrate its utility by comparing beliefs about three types of symptom (respiratory, musculoskeletal and gastrointestinal). Interviews of 150 patients generated items for the belief questionnaire which was then completed by a second sample of 406 general practice patients. Principal components analysis of the responses identified eight readily interpretable belief dimensions: stress; lifestyle; wearing out; environment; internal-structural; internal-functional; weak constitution; concern. Scales were constructed to measure each dimension and the symptom groups were compared. Gastrointestinal symptoms were the most likely to be attributed to internal malfunction and to lifestyle or weak constitution. Musculoskeletal symptoms were more likely to be attributed to structural problems caused by the body wearing out and respiratory symptoms, in contrast, to the influence of the environment. Contrary to prediction, attribution to stress was made equally for the different types of symptom. We have devised a questionnaire, valid specifically for general practice patients, which permits the quantification of beliefs in this setting. The questionnaire could be used in future to track how beliefs respond to medical intervention and how, in turn, beliefs influence illness behaviour.