Three double blind, placebo controlled, cross-over studies were conducted to assess the activity of analgesics in healthy volunteers using a modified submaximal effort tourniquet test (SETT). On each study day tests were performed in a silent cubicle at 21 degrees C immediately before and then hourly after drug administration. After exsanguination of the dominant arm trained subjects performed intermittent, isometric hand-grip exercise for 1 min and then rated pain intensity continuously on a visual analogue scale generated and scored by microcomputer. Tests were terminated by the subjects when pain became unbearable. Tourniquet times and cumulative analgesia scores were significantly increased by small doses of opiates, dipipanone and codeine but not by aspirin, indomethacin or diazepam. Scores on visual analogues scales for alertness and reports of side effects as well as the absence of analgesic activity of diazepam showed that analgesia was not related to sedation. The computerisation of pain ratings and standardisation of experimental conditions probably account for the sensitivity of this model. The difference between responses to opiates and anti-inflammatory drugs is discussed.