1. Experiments were performed to check the validity of previous claims concerning the ;long-loop' aetiology of ;late' reflex electromyogram (e.m.g.) responses to muscle stretch in man. The primary aim was to investigate whether observations previously presented in favour of the ;long-loop hypothesis' are explicable also in terms of the ;resonance hypothesis', according to which the ;late' reflex components represent spinal, short-latency responses to intramuscular oscillations initiated by the impact.2. The contracting wrist flexors of healthy subjects were exposed to trains of recurrent 25-50 Hz stretch stimuli (wrist torque pulses). Each of the initial two or three pulses in the train was followed by e.m.g. peaks with a latency of 20-25 msec. The e.m.g. peaks driven in this way had the following characteristics in common with the successive two or three e.m.g. peaks which were induced by single ramp stretches or tendon taps. (a) Changes in stimulus parameters which altered the strength of the initial e.m.g. peak often had an opposite effect on the strength of the succeeding peak(s). Muscle vibration which attenuated the initial peak often enchanced the succeeding one(s). (b) The initial e.m.g. peak was less affected than the succeeding peak(s) by the subjects' attempts to respond with rapid ;resist' or ;let go' reactions.3. Intramuscular oscillations (monitored by a needle accelerometer) and e.m.g. responses evoked by single ramp stretches and/or tendon taps were also studied in the long thumb flexor, the calf muscles and the masseter muscle. In the thumb flexor, the initial accelerometer deflexion was only rarely succeeded by a short latency e.m.g. peak, but the succeeding wave in the needle accelerogram was followed by such a peak, appearing about 40 msec after stimulus application. By contrast, the calf muscles and the jaw elevators exhibited a high amplitude, short-latency e.m.g. response to the first but only rarely to the second intramuscular oscillation wave.4. The interval between initial and second e.m.g. peaks following tendon taps was longer for calf muscles than for wrist flexors and longer for wrist flexors than for jaw elevators. Similar differences were observed with respect to the intervals between the damped intramuscular oscillations initiated by the impacts.5. Without denying the existence of ;long-loop reflexes' it is concluded that the characteristics of the ;late' reflex responses to muscle stretch in man are explicable also in terms of the ;resonance hypothesis'.