A comparison of intra-articular methotrexate and intra-articular triamcinolone hexacetonide was made in 42 arthritic patients with persistent bilateral knee effusions. One knee was injected with either 5 mg methotrexate (two injections of 2.5 mg a week apart) or a single injection of 20 mg triamcinolone. An objective assessment of both knees was made by quantitative thermography at 0,3,7,14 and 21 days. Joints injected with triamcinolone showed a greater fall in thermographic index (T.I) than the joints injected with methotrexate, which showed similar change to the non-injected knee joints in both groups. Four patients received larger doses of methotrexate, up to 20 mg, though the fall in T.I. was still less than the mean fall for triamcinolone injected joints. Peak venous blood levels of methotrexate were reached 1 hour after intra-articular injection, and a sphygmomanometer cuff inflated around the leg above the injected knee for periods of up to 1 hour did not appreciably delay this. Methotrexate had no immediate anti-inflammatory effect, even in psoriatic arthropathy, and did not give the relief of intra-articular steroid.