Background: In an attempt to reduce heavy lifting exposures, the manual materials handling burden has shifted towards pushing and pulling. Pushing and pulling may pose a biomechanical risk due to excessive loads placed onto the lumbar spine, particularly in anterior/posterior (A/P) shear (Knapik and Marras 2009). The only risk limits available in the scientific literature for pushing and pulling were psychophysically-determined, relying on the assumption that subjective perception of an individual’s maximum acceptable external forces corresponds to biomechanical tolerance (Snook and Ciriello 1991). However, individuals are unlikely able to sense biomechanical loading on critical tissues in the spine due to the lack of nociceptors in the intervertebral disc (Adams et al. 1996). As such, the objective of this study was to create a set of biomechanically-determined risk limits for occupational pushing and pulling that are protective of the low back. Methods: Sixty-two subjects (31 male, 31 female) performed occupational pushing and pulling tasks in a laboratory. Subjects performed three types of exertions (one-handed pull, two-handed pull, two-handed push) at three handle heights (32 in., 40 in., 48 in.) and in one of two directions (straight or turn). Subjects pushed or pulled on custom-built hand transducers connected to an overhead braking system via a rig while performing each exertion. To document a wide range of pushing and pulling exposures, the braking system incrementally increased the linear or rotational resistance proportional to the subject’s changes from the initial global position throughout each trial; subjects exerted up to a maximum voluntary exertion. Dependent measures consisted of the magnitude and direction of three-dimensional forces recorded at the hands, turning torques, net joint moments calculated at each shoulder, and three-dimensional spinal loads (compression, A/P shear, lateral shear) at the superior and inferior endplates of each spinal level extending from T12/L1 to L5/S1, as calculated by a dynamic EMG-driven biomechanical spine model (Knapik and Marras 2009; Hwang et al. 2016a; Hwang et al. 2016b). Multiple linear regression techniques correlated spinal loads with hand force or turning torque in order to develop biomechanically-determined hand force and turning torque limits. The values for straight two-handed pushing and pulling were also compared to psychophysically-determined thresholds developed by Snook and Ciriello (1991). Results and Discussion: The independent measures (exertion type, handle height, and exertion direction) and their interactions significantly influenced dependent measures of hand force, turning torque, shoulder moment, and spinal load. In agreement with Knapik and Marras (2009), spinal loads most frequently exceeded tissue tolerance limits for spinal loading (NIOSH 1981; Gallagher and Marras 2012) in A/P shear. The biomechanically-determined limits developed from this work are up to 30% lower than the closest psychophysically-derived equivalents (Snook and Ciriello 1991). Conclusion: Psychophysically-derived hand force limits are not protective enough of biomechanical risk imposed onto the lumbar spine during pushing and pulling. The biomechanically-determined pushing and pulling guidelines proposed herein provide a more objective and conservative indication of risk and should be implemented moving forward.