Stretch tests are attractive in the diagnosis of nerve root or peripheral nerve lesion. Interpretation of the test results is often difficult since the distribution of tensile forces along the nerve caused by the test manoeuvre is not known. In this study the effect on median nerve tension of 22 positions of the arm was measured with 'buckle' force transducers. With the elbow in full extension and the hand in neutral position, altering the position of the shoulder significantly influenced tension in the proximal part of the median nerve; tension in the distal part was not influenced. With the shoulder in 90 degrees abduction, dorsiflexion of the hand combined with an extended elbow resulted in an increased tension in both distal and proximal parts of the median nerve. Dorsiflexion of the hand combined with flexion of the elbow caused an increase in tension only in the distal part. At all sites of the median nerve the median nerve upper limb tension test caused a significantly higher tension than the radial and ulnar nerve upper limb tension tests. This study provides insight in the normal distribution of tensile forces along the median nerve and can have clinical consequences. For differentiating nerve root from peripheral nerve lesions a specific provocative tension test for the median nerve is advocated. The results of this study provide a theoretical basis for differentiating between lesions in the proximal and distal parts of the median nerve. RELEVANCE: In the diagnosis of nerve(root) lesions, tests in which stretching the nerve provokes the symptoms are widely used, but no data are available on the distribution of tensile forces along nerves and nerve roots during such tests. In this study a human cadaver model is presented to analyse tensile force distribution on the median nerve; it can be an aid to evaluate the specificity of a clinical provocation test for the median nerve.