Our aim was to develop a reproducible and simple radiological scoring system for ankylosing spondylitis (AS) to use in cross sectional and prospective studies. Regarding validation of the BASRI (Bath Ankylosing Spondylitis Radiology Index), radiographs of 470 patients with AS were scored using the New York criteria for the sacroiliac joints. The lumbar and cervical spine, and hips were similarly graded 0-4. These scores were added together to give BASRI-t (total) and if the hips are excluded to give BASRI-s (spine). Radiographs of 188 patients were used to test reproducibility. Blinded radiographs of 89 non-AS patients were included randomly to assess disease specificity. Sensitivity to change was assessed using 177 radiographs from 40 patients. Regarding the cross sectional study, 2200 radiographs of 550 (104 F:446 M) patients were randomly selected and scored using BASRI. The frequency distribution of BASRI-t and BASRI-s were plotted using a probit plot. Inter and intraobservation showed between 73 and 82% and 73 and 88% complete agreement, with specificity of 0.78-0.89, suggesting scores are disease-specific. Sensitivity to change became apparent at 2 years (p<0.05). Scoring required 30 seconds to complete. BASRI-t was found to be normally distributed using a probit plot. The mean BASRI scores (total, spinal, hip) increased with disease duration. The correlation, however, was poor (r=0.293, 0.347, 0.263, respectively). Those with hip involvement had more severe spinal disease (p<0.0001). Men had more severe spinal disease than women (p<0.0001). We conclude BASRI is a reliable and rapid method to grade radiographic changes in AS. Using this scoring system it can be seen that AS is a slowly progressive disease with much individual variation. Hip patients have more severe spinal disease than those without hip involvement and men have more severe spinal disease than women.