Different modes of load applications are used to simulate flexion and extension of the upper body. It is not clear which loading modes deliver realistic results and allow the comparison of different studies. In a numerical study, a validated finite element model of the lumbar spine, ranging from the vertebra L1 to the disc L5-S1 was employed. Each of six different loading modes was studied for simulating flexion and extension, including pure moments, an eccentric axial force, using a wedged fixture, and applying upper body weight plus follower load plus muscle forces. Intersegmental rotations, intradiscal pressures and facet joint contact forces were calculated. Where possible, results were compared to data measured in vivo. The results of the loading modes studied show a large variance for some values. Outcome measures such as flexion angle and intradiscal pressure differed at a segment by up to 44% and 88%, respectively, related to their maximum values. Intradiscal pressure is mainly determined by the magnitude of the applied compressive force. For flexion maximum contact forces between 0 and 69 N are predicted in each facet joint for different loading modes. For both flexion and extension, applying upper body weight plus follower load plus muscle forces as well as a follower load together with a bending moment delivers results which agreed well with in vivo data from the literature. Choosing an adequate loading mode is important in spine biomechanics when realistic results are required for intersegmental rotations, intradiscal pressure and facet joint contact forces. Only then will results of different studies be comparable.