Intraneural microstimulation (i.n.m.s.) was performed in awake human volunteers, using tungsten micro-electrodes inserted into median and ulnar nerve fascicles supplying the skin of the hand. The same electrodes were used alternatively to record impulse activity from single nerve fibres at the i.n.m.s. sites. Monitoring occasionally, with a proximal electrode, the impulse traffic evoked by i.n.m.s. distally in the same fascicle, established that the stimulation procedure could be made selective enough to activate single myelinated fibres in isolation, while also permitting multifibre recruitment. Monitoring propagated impulses also established that i.n.m.s. of a single myelinated fibre supplying a low-threshold mechanoreceptor in the hand might evoke an elementary sensation. Such sensations were fully endowed with cognitive attributes amenable to psychophysical estimation: quality, magnitude and localized projection. Psychophysical tests were made during i.n.m.s. at intraneural sites where single-unit activity was recorded from classified RA, PC, SA I or SA II mechanoreceptors. Changes in excitability of the nerve fibre of an identified unit, induced by further i.n.m.s., certified that the recorded unit had been stimulated during psychophysical tests. Comparing physical location of the receptive field of a recorded unit and localization of the projected field of the corresponding elementary sensation, revealed that either predicted the other accurately. This further assisted identification of the unit activated by i.n.m.s. The type of a recorded unit and the quality of the elementary sensation evoked by its activation were also reciprocally predictive. RA units evoked intermittent tapping, PC units vibration or tickle and SA I units evoked pressure. SA II units evoked no sensation when activated in isolation. Afferent impulse frequency determined magnitude of pressure in the SA I system, and frequency of vibration in the PC system. In the RA system, stimulation frequency determined frequency of tapping-flutter-vibration sensation at relatively low rates and subjective intensity of sensation at high rates. These findings endorse the concept that quality of sensation is coded in specific sensory systems. Further, they provide novel evidence that sensory quality, magnitude and localization can be exquisitely resolved at cognitive levels on the basis of input initiated in a single mechanoreceptor unit.