A series of experiments has been carried out with single doses of simple analgesics in patients with rheumatoid arthritis using a consistent polyad design. This method proved to be both valid and useful. Pain relief scores were a better measure of the effectiveness of analgesics than preference.Aspirin, Codis, and Distalgesic were the most effective analgesics tested, with paracetamol, pentazocine, and Ciba 44,328 intermediate between these agents and placebo. Placebo given after an active analgesic was more effective than when given before; this phenomenon was not abolished by telling the patients that apparently identical tablets were, in fact, different or by making them different in colour. The effectiveness of soluble placebo depended on its colour, red being the most effective.