Single-pulse Transcranial Magnetic Stimulation (TMS) and, very recently, lumbar stimulation (LS) have been used to measure cortico-spinal excitability from various interventions using maximal or submaximal contractions in the lower limbs. However, reliability studies have overlooked a wide range of contraction intensities for MEPs, and no reliability data is available for LEPs. This study investigated the reliability of motor evoked potentials and lumbar evoked potentials at different stimulation intensities and contraction levels in m.rectus femoris. Twenty-two participants performed non-fatiguing isometric knee extensions at 20 and 60% of maximum voluntary contraction (MVC). LS induced a lumbar-evoked potential (LEP) of 25 and 50% resting maximal compound action potential (M-max). TMS stimulator output was adjusted to 120, 140, and 160% of active motor threshold (aMT). In each contraction, a single MEP or LEP was delivered. Ten contractions were performed at each stimulator intensity and contraction level in random order. Moderate-to-good reliability was found when LEP was normalized to M-max/Root Mean Square in all conditions (ICC:0.74–0.85). Excellent reliability was found when MEP was normalized to Mmax for all conditions (ICC > 0.90) at 60% of MVC. Good reliability was found for the rest of the TMS conditions. Moderate-to-good reliability was found for silent period (SP) elicited by LS (ICC: 0.71–0.83). Good-to-excellent reliability was found for SP elicited by TMS (ICC > 0.82). MEPs and LEPs elicited in m.rectus femoris appear to be reliable to assess changes at different segments of the cortico-spinal tract during different contraction levels and stimulator output intensities. Furthermore, the TMS- and LS- elicited SP was a reliable tool considered to reflect inhibitory processes at spinal and cortical levels.