PURPOSE: Lateral epicondylitis is a painful condition affecting the proximal enthesis of the extensor carpi radialis brevis tendon. Although magnetic resonance imaging (MRI) has been highlighted as an important diagnostic tool, to our knowledge no previous study has established the observer reliability of MRI for lateral epicondylitis or the relationship between MRI abnormalities of the common extensor origin and the patient's clinical assessment. METHODS: Twenty-one consecutive subjects with a clinical diagnosis of chronic lateral epicondylitis were assessed. An MRI scoring system was used to grade the degree of tendinosis and length of tendon separation of the common extensor origin from the lateral epicondyle. Three independent musculoskeletal radiologists, who were blinded to patient clinical severity, scored images separately. Each scored the images on 3 separate occasions. Clinical symptoms were assessed using the Quick Disabilities of the Arm, Shoulder, and Hand (QuickDASH) and Upper Extremity Functional Scale clinical measures. Maximum pain levels were scored on a visual analog scale, and objective assessment was made with grip strength. RESULTS: Moderate or severe signal changes consistent with tendinosis were observed in 18 of 21 patients. Significant inter-observer reliability and intra-observer agreement were demonstrated for MRI interpretation of grade of tendinosis and length of tendon separation. Significant negative correlation was found between the length of tendon separation and both the QuickDASH and maximum pain levels. CONCLUSIONS: Magnetic resonance imaging is a reliable tool in determining radiological severity of lateral epicondylitis. However, the severity of MRI signal changes does not positively correlate with symptoms. These findings question the validity of MRI in the assessment of lateral epicondylitis.