Percutaneous spinal stimulation is a promising new technique for understanding human spinal reflexes and for evaluating the pathophysiology of motor roots. Previous studies have generally stimulated the T11/T12 or T12/L1 vertebral junctions, sites that overlie the lumbosacral enlargement. The present study sought to determine the best location for targeting sensory and motor roots during sitting. We used paired stimuli, 50 ms apart, to distinguish the contribution of the reflex and motor components which make up the root evoked potential. This assumed that post-stimulation attenuation, primarily through homosynaptic depression, would abolish the second potential if it was trans-synaptic in origin. Conversely, successive responses would be unchanged if motor roots were being stimulated. Here, we show that sensory root reflexes were optimally elicited with percutaneous stimulation over the L1-L3 vertebrae. However, the optimal position varied between subjects and depended on the target muscle being studied. A collision test showed that the reflex recorded in pre-tibial flexors was low in amplitude and was prone to crosstalk from neighbouring muscles. In contrast to the reflex response, direct motor root activation was optimal with stimulation over the more caudal L5-S1 vertebrae. The present results support the utility of paired stimulation for evaluating the topographical recruitment of sensory and motor roots to human leg muscles.