Ipsilateral knee (iKnee) joint rotations in seated humans elicit short-latency crossed spinal reflexes in the contralateral biceps femoris (cBF) muscle (Stevenson et al., 2012). The short-latency cBF reflexes were inhibitory following iKnee extension perturbations, and facilitatory following flexion perturbations. Due to the onset latency (45 ms), spinal pathways likely mediate the reflexes. In the present study, intramuscular electromyography (iEMG) was recorded in the cBF to determine whether the cBF reflexes can be quantified at the motor unit (MU) level and involved the same population of MUs. 11 seated participants (mean age 25 ± 5 years) performed a voluntary isometric knee extension with the ipsilateral leg and contralateral knee flexion to 10% of MVC. Surface EMG was recorded bilaterally from the BF and rectus femoris, and iEMG from the cBF. A mechanical actuator (MTS-Systems Corporation) imposed iKnee flexion or extension perturbations (8° and 150°/s) in blocks of 60 trials. iEMG data for flexion and extension perturbations were decomposed (EMGLAB, McGill et al., 2005) into constituent MU action potentials (APs). The total number of APs was quantified using a 10 ms window to create peristimulus time histograms (PSTHs). Periods of facilitation and inhibition were determined if the firing counts in three or more adjacent bins were above (or below) the mean background firing plus (or minus) three standard deviations (Mao et al., 1984). Across all participants, an inhibitory cBF reflex was observed at a latency of 40-50 ms after the onset of iKnee extension perturbations, lasting for 30-40 ms. Following iKnee flexion perturbations, a facilitatory cBF reflex was observed with an onset latency of 40-50 ms, lasting for 60-70 ms. The background MU firing frequencies in the cBF were 12.4 ± 3.2 pps and 13.5 ± 4.3 pps during flexion and extension perturbations, respectively. The mean background firing frequency was not significantly different between perturbation directions, p = .2. A two-way repeated measures ANOVA revealed a significant interaction between perturbation direction and time window (background, response) F(1,10) = 48.8, p