A questionnaire, which allowed graphing of self-assessment of active and passive shoulder motion by drawing lines onto prepared diagrams, was sent to 221 consecutive patients scheduled to attend our outpatient clinic. At consultation, shoulder motion was measured using a goniometer by surgeons, who were blinded to the results of the patient-based self-assessment. One-hundred and fifty-eight complete data sets were available for evaluation. Mean differences between patient and surgeon measurements were 18 degrees +/- 19 degrees . While mean values for most degrees of freedom were similar between patient and surgeon measurements, some rotational movements were overestimated by patients. Correlation of patient and surgeon based assessments were poor for all degrees of freedom (r(2) <or= .52). Dominance (right/left handed), gender, age, pathology, pain, and subjective shoulder value did not significantly impact on the correlation. Combining surgeon and patient based information in the same study is inappropriate. Comparisons and meta-analysis of shoulder studies need to take the type of data acquisition into account.