OBJECTIVES This study aimed to describe the epidemiology of ankylosing spondylitis (AS) in rheumatology practice at the beginning of the anti-TNF (tumour necrosis factor) era, and to evaluate the initiation of anti-TNF therapy in a clinical setting where prescription is regulated by the authority's imposed reimbursement criteria. METHODS Between February 2004 and February 2005, all Belgian rheumatologists in academic and non-academic outpatient settings were invited to register all AS patients who visited their practice. A random sample of these patients was further examined by an in-depth clinical profile. In a follow-up investigation, we recorded whether patients initiated anti-TNF therapy and compared this to their eligibility at baseline evaluation. RESULTS 89 rheumatologists participated and registered 2141 patients; 1023 patients were clinically evaluated. These 847 fulfilled the New York modified criteria for definite AS and 176 for probable AS. The profile of AS in rheumatology practice is characterised by longstanding and active disease with a high frequency of extra-articular manifestations and metrological and functional impairment. At a median of 2 months after the clinical evaluation, anti-TNF therapy was initiated in 263 of 603 (44%) evaluable patients with definite AS and in 22 of 138 (16%) evaluable patients with probable AS (total 38%). More than 85% of the patients who started anti-TNF therapy had an increased Bath Ankylosing Spondylitis Disease Activity Index despite previous NSAID (non-steroidal anti-inflammatory drug) use. CONCLUSIONS Of a representative cohort of 1023 Belgian AS patients seen in daily rheumatology practice, about 40% commenced anti-TNF therapy. Decision factors to start anti-TNF therapy may include disease activity and severity.