1. Microneurographic techniques were used to isolate single afferent axons within cutaneous and motor fascicles of the median and ulnar nerves at the wrist in thirteen subjects. Of the sixty-five identified afferents, eleven innervated the interphalangeal and metacarpophalangeal joints, sixteen innervated muscle spindles, three innervated Golgi tendon organs and thirty-five supplied the glabrous skin of the hand. 2. Intrafascicular stimulation through the recording microelectrode, using trains of constant-voltage positive pulses (0.3-0.8 V, 0.1-0.2 ms, 1-100 Hz) or constant-current biphasic pulses (0.4-13.0 microA, 0.2 ms, 1-100 Hz), evoked specific sensations from sites associated with some afferent species but not others. 3. Microstimulation of eight of the eleven joint afferent sites (73%) evoked specific sensations. With four, subjects reported innocuous deep sensations referred to the relevant joint. With the other four, the subjects reported a sensation of joint displacement that partially reflected the responsiveness of the afferents to joint rotation. 4. Microstimulation of fourteen of the sixteen muscle spindle afferent sites (88%) generated no perceptions when the stimuli did not produce overt movement. However, subjects could correctly detect the slight movements generated when the stimuli excited the motor axons to the parent muscle. 5. With seven of the nine rapidly adapting (type RA or FAI) cutaneous afferents (88%) microstimulation evoked sensations of 'flutter-vibration', and with two of eight slowly adapting (type SAI) afferents (25%) it evoked sensations of 'sustained pressure'. Of the eighteen SAII afferents, which were classified as such by their responses to planar skin stretch, the majority (83%) generated no perceptions, confirming previous work, but three evoked sensations of movements or pressure. 6. The present results suggest a relatively secure transmission of joint afferent traffic to perceptual levels, and it is concluded that the human brain may be able to synthesize meaningful information on joint displacement on the basis of impulses in a single joint afferent. This could partly compensate for the low responsiveness of individual joint afferents within the physiological range of joint displacements. Although single muscle spindle afferents can adequately encode joint position and movement, the results suggest that the brain needs the information from more than one muscle spindle afferent to perceive changes in joint angle.