Objectives: To determine the type and proportion of patients with ankylosing spondylitis who rheumatologists consider to be candidates for treatment with tumour necrosis factor (TNF)-blocking agents, and to what extent this is in agreement with the ASsessment in Ankylosing Spondylitis (ASAS) international working group recommendations on initiation of treatment with anti-TNF agents. Methods: Participants were rheumatologists from 10 different countries, who were considered to be experts in treating patients with ankylosing spondylitis and in the use of anti-TNF treatment, but were unaware of the ASAS recommendations (unpublished at the time of study in 2003). The first 10 consecutive patients with ankylosing spondylitis seen by the rheumatologist were evaluated as to whether the patient was a candidate for anti-TNF treatment. Thereafter, a metrologist assessed the patient for disease activity and severity, and collected data on demographics and treatment. Results: Complete data were available for 1207 of the 1284 patients and were used for analysis. Overall, the rheumatologists indicated that they would initiate TNF-blocking agents in 49.3% of patients, ranging from 37.2% patients in Canada to 78.3% in Australia. These candidates had higher disease activity, higher levels of acute-phase reactants, worse spinal mobility, worse function, more often hip involvement and a higher prevalence of sick leave. Of all patients considered to be candidates, 40% did not fulfil ASAS recommendations with respect to previous use of non-steroidal anti-inflammatory drugs (NSAIDs; at least two NSAIDs) or Bath Ankylosing Spondylitis Disease Activity Index (>= 4). Conversely, 36% of patients who did not fulfil the NSAID or BASDAI recommendations were still considered to be candidates for TNF-blocking treatment. Objective variables, such as C reactive protein, erythrocyte sedimentation rate or magnetic resonance activity, were considered less important than disease activity in the decision on starting TNF-blocking drugs. The only important objective criterion was rapid radiographic progression. Conclusion: Rheumatologists wanted to initiate TNF-blocking drugs in roughly half of the patients with ankylosing spondylitis. However, there was a wide variation across countries and doctors. Rheumatologists considered both disease activity and severity to be determinants of starting TNF blockers, but their decision was often in disagreement with ASAS recommendations.