The effects of a voluntary contraction on the H reflexes of various muscles were quantified to determine whether the reflex responses were sufficiently reproducible to be used in diagnostic studies. During a voluntary contraction, H reflexes could be recorded reliably from tibialis anterior and abductor pollicis brevis, but accurate identification of the onset of the H wave from the on-going background EMG required duplicate averages of multiple responses. During a contraction the H reflex could be obtained at lower stimulus intensities in the forearm flexor muscles than when relaxed, and a more clear separation of the H wave from the M wave was possible. The background contraction abolished the attenuation of reflex amplitude with increasing stimulus repetition rates, such that repetition rates of up to 4 Hz could be used without significant loss of reflex amplitude. There were only small and usually insignificant differences in the latency of the H reflex or its variability when elicited with the forearm muscles relaxed and when flexor carpi radialis was contracting steadily. The reflex latencies of abductor pollicis brevis, tibialis anterior and soleus were compared with F wave latencies for these muscles. The minimal F wave latencies were shorter than the H reflex latencies for abductor pollicis brevis (mean 2.2 ms) and tibialis anterior (mean 1.0 ms) but not for soleus Comparison of the spread of F wave latencies (F max-F min) suggests that, for soleus, F waves are recorded only from the faster conducting motor units in the pool, presumably those less readily recruited in the H reflex It was calculated that the distribution of motor conduction velocities responsible for the F waves of abductor pollicis brevis was 8.8 m.s-1 This value underestimates the likely distribution of motor conduction velocities for the thenar muscle by as much as 50%, consistent with the view that F waves rarely occur in slowly conducting motor units, the units of lowest threshold in reflex studies It is concluded that, for many motoneuron pools, the H reflex and the F wave appear preferentially in different motoneurons, low and high threshold, respectively, and that reflex studies can provide information not available from somatosensory evoked potentials or F wave studies.