Drug studies in ankylosing spondylitis (AS) are usually short term, in highly selected patients at academic centers. We present data on 1331 UK patients with AS. Patients were reviewed prospectively in 1985 and 1987. Given the nationwide population base we avoided biases relating to local medical preferences or market forces. Eighty-six percent (n = 1149) were taking medication in 1985 and 78% (n = 1040) in 1987. The most common drug in 1985 was indomethacin (Indo) with 35%, followed by naproxen (N; 21%), piroxicam (P; 9%), diclofenac (D; 7%) and ibuprofen (6%; all others less than 5%). Two years later the figures were Indo 34%, N 19% and D 12%. At followup, survival rates (the number remaining taking each drug) were Indo 75%, N 63%, P 74% and D 67%. The 2 main reasons for stopping a drug were lack of efficacy (Indo 10%, N 25%, P 34%, D 32%) and toxicity (Indo 39%, N 30%, P 30%, D 40%). Sixty percent of patients taking Indo reported good or excellent pain relief, compared with 57% on N, 47% on P and 47% on D. Good or excellent stiffness relief was obtained in 55% of patients on Indo, 38% on N, 44% on P and 40% on D. At any one time over 75% of individuals with AS are receiving drug treatment. The most popular nonsteroidal antiinflammatory drug for AS in 1985 and 1987 was Indo, which scored highest in terms of efficacy, pain relief, stiffness relief and survival. The survival rate over 2 years ranged from 63% (N) to 75% (Indo).