Shoulder pain is a common musculoskeletal condition and affects a large proportion of the UK population. Over half of those affected continue to have pain eighteen months from onset. In order to help with the diagnosis and management of shoulder pain, the use of ultrasound scans has been increasing. Despite this rise in ultrasound scans, the predictive value of ultrasound-based pathology findings with outcomes remains unclear. Not all patients in pain necessarily require investigations or treatment. Evaluating the relationship between pain and an acceptable symptom state is also important to understand which patients require interventions. The over-arching hypothesis underlying this thesis was that aspects of the shoulder pain pathway can be improved through better utilisation of ultrasound and applying the concept of a patient acceptable symptom state. A systematic literature review on the role of imaging and shoulder symptoms identified a paucity of studies evaluating multiple concurrent imaging pathologies with shoulder symptoms. A retrospective study using latent class analysis demonstrated that groups of ultrasound-detected pathologies existed. A prospective study confirmed the existence of these groups. However, there was no difference in 6 months outcome or response to treatments between these groups or individual pathologies. Patients with worse symptoms at baseline were more likely to find worse symptoms acceptable at 6 months. Patients who reported acceptable symptoms at baseline also received fewer treatments. In summary, the current use of ultrasound scans in managing patients with shoulder pain needs re-evaluation, and understanding clinical criteria such as the patient acceptable symptom state will help improve shoulder pathways. These findings should inform future trial designs and shoulder care pathways.