The introduction of symptomatically highly effective anti-tumour necrosis factor alpha therapies for ankylosing spondylitis (AS) has generated interest in the use of imaging to evaluate the potential structure-modifying properties of these agents. Several approaches have been developed to score the plain radiographic abnormalities in AS. Of these, the modified Stoke AS Spinal Score is the most responsive to change, although responsiveness is limited and requires a minimum of 2 years before significant change becomes apparent in patients on standard therapies. Magnetic resonance imaging (MRI) is the most sensitive imaging abnormality, and the advent of fat-suppression imaging allows detection of bone marrow inflammation in the sacroiliac joints as one of the earliest abnormalities in AS. Limited studies have shown that spinal inflammation can be scored reliably using either a system that evaluates the entire spine or a system that limits evaluation to only the most severely affected spinal segments. Both methods also demonstrate excellent responsiveness. The prognostic significance of acute changes on MRI remains unclear. Reliable approaches to the evaluation of chronic changes are yet to be developed. MRI represents a major advance in the diagnostic evaluation of AS.