Patients with sensory disturbances of painful and non-painful character show distinct changes in touch and/or pain sensitivity. The patterns of sensory changes were compared to those of human surrogate models of neuropathic pain to assess the underlying mechanisms. We investigated 30 consecutive in-patients with dysaesthesia of various origins (peripheral, spinal, and brainstem lesions) and 15 healthy subjects. Tactile thresholds were determined with calibrated von Frey hairs (1.1mm). Thresholds and stimulus-response functions for pricking pain were determined with a series of calibrated punctate mechanical stimulators (0.2mm). Allodynia was tested by light stroking with a brush, Q-tip, and cotton wisp. Perceptual wind-up was tested by trains of punctate stimuli at 0.2 or 1Hz. Intradermal injection of capsaicin (n=7) and A-fiber conduction blockade (n=8) served as human surrogate models for neurogenic hyperalgesia and partial nociceptive deafferentation, respectively. Patients without pain (18/30) showed a continuous distribution of threshold shifts in the dysaesthetic skin area with a low to moderate increase in pain threshold (by 1.52+/-0.45 log2 units). Patients with painful dysaesthesia presented as two separate groups (six patients each): one showing lowered pain thresholds (by -1.94+/-0.46 log2 units, hyperalgesia) and the other elevated pain thresholds (by 3.02+/-0.48 log2 units, hypoalgesia). The human surrogate model of neurogenic hyperalgesia revealed nearly identical leftward shifts in stimulus-response function for pricking pain as patients with spontaneous pain and hyperalgesia (by a factor of about 5 each). The sensory changes in the human surrogate model of deafferentation were similar to patients with hypoalgesia and spontaneous pain (rightward shift of the stimulus-response function with a decrease in slope). Perceptual wind-up did not differ between symptomatic and control areas. There was no exclusive association of any parameter obtained by quantitative sensory testing with a particular disease (of either peripheral or central origin). Our findings suggest that neuropathic pain is based on two distinct mechanisms: (I) central sensitization (neurogenic hyperalgesia; in patients with minor sensory impairment) and (II) partial nociceptive deafferentation (painful hypoalgesia; in patients with major sensory deficit). This distinction as previously postulated for postherpetic neuralgia, is obviously valid also for other conditions. Our findings emphasize the significance of a mechanism-based classification of neuropathic pain.