The concept of disease modification in ankylosing spondylitis (AS) incorporates aspects of inflammation, bone destruction and new bone formation. The degree to which inflammation and new bone formation are linked remains conjectural, but data from MRI studies of spinal inflammation support the concept of such coupling; however, these studies also suggest a role for the involvement of noninflammatory pathways, such as those involving bone morphogenetic proteins, wingless proteins and Dickkopf-1, in the formation of new bone. The main clinical outcome that reflects disease modification is the modified Stoke Ankylosing Spondylitis Spine Score, which assesses abnormalities in the anterior vertebral corners of the cervical and lumbar spine. However, radiographic progression can only be reliably detected using this method after at least 2 years, and this delay precludes the conduct of placebo-controlled trials on ethical grounds. Preliminary data using this scoring tool suggest that cyclooxygenase-2-selective NSAIDs might reduce disease progression if used continuously over 2 years. By contrast, three different anti-tumor necrosis factor therapies have shown no impact on radiographic progression. Therapeutic trials recruiting patients early in their disease course and at high risk of radiographic progression constitute a high priority for clinical research in AS.