1. Different techniques were used to generate sudden ramp extension movements of the wrist while the subjects were either relaxed or maintaining a weak voluntary contraction in the wrist flexors. Afferent responses to the displacements were recorded with a tungsten micro-electrode inserted into a fascicle of the median nerve supplying one of the wrist flexor muscles, and e.m.g. responses were recorded with needle electrodes inserted into the same muscle.2. With the wrist flexors either relaxed or contracting, extensions at 100-200 degrees /sec for 60-70 msec (generated by either an hydraulic motor or a torque motor) produced segmented afferent responses with two to four afferent bursts, separated by intervals of 20-30 msec. The successive neural peaks, occuring during the stretch phase, were correlated to mechanical vibrations sensed by a strain gauge and sometimes also by a wrist goniometer. With the flexor muscles contracting, the successive peaks in the neurogram were followed by similar peaks in the e.m.g, the delay between neural and e.m.g. peaks being 20-25 msec.3. Small abrupt extension movements of 1-2 degrees lasting only 10-15 msec often produced segmented afferent responses with one neural burst occuring during the stretch phase and another 15-20 msec later, corresponding to a mechanical oscillatory event succeeding the stretch. The oscillation and the second neural burst were not present with small extension movements of smooth onset and halt. With the flexor muscles contracting, stimuli producing one afferent burst produced only one e.m.g. peak, whereas double-peaked afferent discharges produced double-peaked e.m.g. responses, the delay between individual neural e.m.g. peaks being 20-25 msec.4. Similar segmentation of the neural stretch responses was seen when abrupt displacements were produced by electrically induced muscle twitches, by manual pulls on a spring attached to the hand or by the subject making fast voluntary wrist extensions. This grouping of afferent discharges was seen in both multi-unit and in single-unit recordings from fibres identified as group Ia afferents.5. It is concluded that mechanical vibrations in the moving parts are initiated by abrupt joint movements and that these vibrations are sensed by the primary endings. With initial background contraction in the stretched muscles, synchronous volleys of spindle discharges produce, via segmental reflex arcs, modulation of the e.m.g. with the appearance of two or three e.m.g. peaks separated by intervals of 20-30 msec. Possible causes for the mechanical oscillations are discussed.6. For imposed movements with a duration of 60-70 msec the successive e.m.g. peaks caused a fused reflex contraction, appearing as a torque trace deflexion, starting at about the time when the movement ended and reaching its peak within about 40 msec. With longer-lasting movements the mechanical reflex response accompanying the successive e.m.g. bursts, appeared as a decelerative force, starting to oppose the ongoing movement about 60 msec after its start. Mechanical consequences of stretch reflex contractions starting after, rather than during, the stretch movement are discussed.