A double-blind crossover trial compared Butacote, 200 mg b.d. with naproxen, 250 mg b.d., each given for four weeks, in the treatment of rheumatoid arthritis. The trial was multicentre, and twenty-six general practitioners admitted forty-eight patients. Seven patients dropped out: two for technical reasons; one in each treatment group because of exacerbation of symptoms; one because of intolerance of rescue analgesic; and two with gastric intolerance to naproxen. Both drugs were effective in relieving pain, morning stiffness, and joint tenderness, as compared with the condition before the trial. There was little effect on grip strength or joint size, and no real differences between the two drugs. The two preparations had a similar frequency and spectrum of side-effects, except that oedema did not occur with naproxen, and rash did not occur with Butacote. Gastro-intestinal upsets were the commonest unwanted effect but only two patients had to stop treatment for this reason, both while taking naproxen. Although the doctors preferred each drug on an equal number of occasions, the patients preferred Butacote to naproxen almost twice as often (20:11), particularly when Butacote was given after naproxen. The reason for this divergence of opinions is not obvious.