Brush-evoked pain (mechanical allodynia, dynamic mechanical hyperalgesia) is a hallmark of neuropathic and inflammatory pain states. Here we have examined the neural mechanisms that induce and maintain this component of mechanical hyperalgesia. The principle finding of these experiments is that the severity of brush-evoked pain correlates with the intensity of background pain in patients suffering from chronic painful neuropathies and in normal subjects with acute experimental chemogenic pain. In experiments on nine normal subjects topical application of mustard oil for 5 min evoked strong burning pain and hyperalgesia to light mechanical stimuli. Differential nerve blocks (by compression of the superficial radial nerve) revealed that the brush-evoked pain was transmitted by A beta-fibres, which normally encode non-painful tactile sensations, while the burning pain was signalled by C-fibres. Psychophysical measurements showed that mustard oil treatment resulted in a pronounced sensitization of nociceptors to heat so that subsequent innocuous changes of skin temperature from 35 to 40 degrees C resulted in a proportional increase of burning background pain. Changes in the magnitude of ongoing burning pain were closely correlated (r = 0.81) to the intensity of brush-evoked pain. While conduction block of A-fibres eliminated only touch-evoked pain, blockade of C-fibre excitation instantaneously abolished both ongoing and touch-evoked pain. In nine patients with chronic neuralgia (15 years mean duration) ongoing and brush-evoked pain were examined. In six patients, differential block of A beta-fibres eliminated touch-evoked pain, but ongoing pain persisted when only C-fibres were conducting. Complete relief of both ongoing and stimulus-induced pain was obtained in two patients with intravenous regional guanethedine block and in two other individuals by local anaesthetic blocks of nerves supplying the symptomatic skin, indicating that input from primary afferents was necessary for the maintenance of the pains and that ongoing pain was not self-perpetuated by central mechanisms alone. Quantitative sensory tests revealed heat hyperalgesia in four patients. In those individuals, an increase of skin temperature produced a graded increase of their ongoing pain which was closely correlated (r = 0.94) with the level of brush-evoked pain. In the remaining five patients there was no heat hyperalgesia and consequently no aggravation of pain by increases of skin temperature. Nevertheless when the intensity of the background pain fluctuated spontaneously there were also parallel changes (r = 0.88) of the severity of brush-evoked pain.(ABSTRACT TRUNCATED AT 400 WORDS)