Recent evidence suggests that shoulder impingement syndrome arises from primary rotator cuff pathology and may be related to the inability of the rotator cuff to prevent superior humeral head migration in shoulder elevation. Impingement involves compression of subacromial structures, including the rotator cuff. Previously, clinical tests have been shown to be inaccurate in diagnosing rotator cuff impingement. A lack of anatomical validity might explain the inaccuracy of these tests. This study aimed to clarify the anatomical basis of subacromial compression of the rotator cuff by analysing the compression forces generated and observing the structures impinged in a variety of shoulder positions.This observational case series involved the dissection of nine embalmed cadaveric shoulders.Pressure transducers were placed deep to the coracoid process, coracoacromial ligament, the anterior acromion and the posterior acromion. Shoulders were moved into internal and external rotation from the positions of flexion, abduction and extension. At each position, pressure readings were recorded and structures being compressed observed visually.Highest pressures were recorded in flexion/internal rotation at the coracoacromial ligament, in abduction/internal rotation at the coracoid process (both involving the rotator interval) and in abduction/internal rotation at the coracoacromial ligament (involving supraspinatus). Supraspinatus was also observed to be compressed in extension/external rotation (against the anterior acromion). Infraspinatus was compressed in extension/external rotation (against the posterior acromion), while subscapularis was compressed in flexion/internal rotation and flexion/external rotation (both against the coracoid process).This study identifies shoulder positions likely to impinge particular rotator cuff tendons.