OBJECTIVE The H-reflex on stimulation of the tibial nerve in the popliteal fossa is routinely used in the diagnosis of first sacral (S1) nerve-root radiculopathy. The H-reflex latency, however, is considered to lack sensitivity since a small change from the focal root pathology can be diluted in a relatively long reflex latency. We have studied the soleus H-reflex elicited by stimulation of the S1 nerve root at the S1 foramen. The normal values for the S1-foramen H-reflex have been reported in a previous study, but there are no definitive reports in patients with S1 radiculopathy. This study was undertaken to determine whether stimulating at the S1 nerve root can improve the utility of the H-reflex for detecting an S1-root lesion. METHODS A randomised paired-study design was utilised to evaluate two H-reflexes: one elicited with tibial nerve stimulation and one elicited with S1-root stimulation. Fifty-five patients with unilateral S1 radiculopathy, confirmed by clinical, electrodiagnostic and magnetic resonance imaging (MRI) evidences were studied. A high-voltage electrical stimulator was used to elicit H-reflexes bilaterally at the S1 foramen and L4/L5 spine level. Latencies were compared with previously generated normal values and similar responses from the asymptomatic leg, focussing on the interval between the peak of M- and H-waves (HMI). RESULTS On the symptomatic side, 39 of the 55 patients had abnormal tibial H-reflex latencies and 54 patients had abnormal responses on S1-foramen stimulation (absent in 18; HMI prolonged >0.4ms in 36). On the asymptomatic side, all 55 patients had normal tibial H-reflexes, and 52 had normal responses on S1-foramen stimulation. In three patients, the HMI was abnormal on S1-foramen stimulation. In 46 patients tested with L4/L5-level stimulation, H-reflex was present in 39 and absent in seven. The latency of the M-wave to S1 stimulation was normal. CONCLUSIONS Abnormal S1-root H-reflexes reveal lesions at the S1 root in patients with normal tibial H-reflexes; therefore, enhancing diagnostic sensitivity. The appearance of the H-reflex to L4/L5-level stimulation in patient with absent H-reflex to S1-foramen stimulation further localises the site of S1 nerve-root lesion to the L5/S1 spine level. Thus, H-reflex to S1-root stimulation significantly increases the diagnostic sensitivity for S1 radiculopathy. CONCLUSIONS In our study, the S1-root H-reflex with high-voltage electrical stimulation has shown greater sensitivity than the tibial H-reflex in evaluating S1 compressive radiculopathies. An abnormal S1-root H-reflex helps to localise the lesion to the S1 root in patients with concurrent abnormal tibial nerve H-reflex, which may increase diagnostic specificity.