We compared the practice of four orthopaedic consultants as regards the use of MRI and arthroscopy to diagnose problems of the knee. In one year 324 arthroscopies and 66 MR scans were performed for this purpose.

We found that MRI is a reliable and cheaper alternative to ‘diagnostic arthroscopy’. We consider that patients with definite clinical signs merit an early ‘therapeutic arthroscopy’, but that all other knees should be investigated by MRI. This policy spares patients from unnecessary and expensive surgery.

Arthroscopy for diagnostic purposes should be used only with a specific purpose. Modern MRI can and should replace “having a look”.

We reviewed 28 patients with brachial plexus lesions caused by shoulder dislocation. Contrary to most other reports, we found that the neurological lesions involved the infraclavicular and the supraclavicular brachial plexus. With supraclavicular lesions the involvement was always of the suprascapular nerve, and this always recovered spontaneously. Isolated axillary nerve lesions had the poorest prognosis for spontaneous recovery. We explored all lesions that showed no recovery after three to five months and performed either grafting or neurolysis. We discuss the combinations of nerve lesions, their recovery, the surgical indications, and the operations. We also suggest a new classification for these injuries which is more clinically relevant than the anatomical classification of Leffert and Seddon (1965).