There are no clear guidelines for diagnostic imaging of articular and soft tissue pathologies of the shoulder and elbow. Several methods are used, including magnetic resonance imaging (MRI), magnetic resonance arthrography (MRA) and ultrasound (US). Their cost-effectiveness is still unclear. We performed a meta-analysis of the relevant literature and discussed the role of MR imaging of the shoulder and elbow compared with other diagnostic imaging modalities. For the shoulder impingement syndrome and rotator cuff tears, MRI and US have a comparable accuracy for detection of full-thickness rotator cuff tears. MRA and US might be more accurate for the detection of partial-thickness tears than MRI. Given the large difference in cost of MR and US, ultrasound may be the most cost-effective diagnostic method for identification of full-thickness tears in a specialist hospital setting (Evidence level 3). Both MRA and CT arthrography (CTA) are effective methods for the detection of labrum tears. More recently, multidetector CTA has offered the advantages of thinner slices than with MRA in a shorter examination time. Still, MRA has the advantage towards CTA to directly visualize the affected structures with a better evaluation of extent and location and to detect associated capsuloligamentous injuries. For the elbow pathologies, plain MRI or MRA have the advantage towards CTA to detect occult bone injuries. CTA is better for the assessment of the thin cartilage of the elbow. Both US and MRI are reliable methods to detect chronic epicondylitis; US is more available and far more cost-effective (Evidence level 2). MRA can differentiate complete from partial tears of the medial collateral ligament. US or MRI can detect partial and complete biceps tendon tears and/or bursitis. MRI can provide important diagnostic information in lesions of the ulnar, radial, or median nerve.