Phenylbutazone in doses of 50, 100, 200 and 400 mg/day has been given for four periods of three weeks to seven patients with rheumatoid arthritis. The trial was double/blind and the order of administration of the doses was arranged to eliminate order and carry-over effects. Before and at the end of each period the patient's clinical state was assessed by measurement of the pain score, duration of morning stiffness, paracetamol tablet count, grip strength, digital joint size, articular index and overall patient preference. The plasma phenylbutazone concentration was measured by gas liquid chromatography and the expected concentration of phenylbutazone in plasma was also predicted from a knowledge of the antipyrine half-life in each patient. Phenylbutazone had a significant therapeutic effect, in some cases after only 50 mg/day. There was however no correlation between the plasma concentration of phenylbutazone and its therapeutic effect. The observed plasma concentration of phenylbutazone agreed well with that predicted at doses of 50 and 100 mg/day. However, at 200 and 300 mg/day the observed plasma concentration was significantly lower than that predicted, perhaps due to saturation of the protein binding sites.