Objective: To investigate the influence of patient characteristics on the course of spinal radiographic progression in a large prospective longitudinal cohort study of ankylosing spondylitis (AS) patients treated long-term with TNF-α inhibitors. Methods: Consecutive patients from the Groningen Leeuwarden AS (GLAS) cohort starting TNF-α inhibitors with spinal radiographs at least available at baseline and 6 years of follow-up were included. Radiographs were scored using mSASSS by two independent readers. Generalized estimating equations (GEE) were used to explore the associations between baseline characteristics and spinal radiographic progression. The course of radiographic progression in patients with and without risk factors for poor radiographic outcome was investigated using different time models (linear and non-linear). Single linear imputation was used in case of missing radiographic data at the intermediate (2 or 4 years) follow-up visits. Results: 80 AS patients were included with mean baseline mSASSS 8.7±13.3. Baseline syndesmophytes, male gender, older age, longer symptom duration, smoking, and higher BMI were significantly associated with more radiographic damage over time. GEE analysis in patients with these risk factors revealed that radiographic progression followed a non-linear course with mean mSASSS progression rates reducing from max. 2.8 units over 0-2 years to min. 0.9 units over 4-6 years. The GEE model revealed a linear course with overall very low progression (≤1 mSASSS units/2yrs) in patients without risk factors. Complete case analysis in 53 patients showed similar results. Conclusion: AS patients at risk of poor radiographic outcome showed the highest but diminishing spinal radiographic progression during long-term treatment with TNF-α inhibitors.