The in vitro multidirectional flexibility analysis was conducted to investigate the initial biomechanical effect of biomimetic artificial intervertebral disc replacement from either anterior or posterior approach in a cadaveric lumbosacral spine model. Two designs of anterior total and posterior subtotal artificial discs were developed using bioactive three-dimensional fabric and bioresorbable hydroxyapatite/poly-l-lactide material (3DF disc). Both models were designed to obtain the stable interface bonding to vertebral endplates with maximum surface area occupation. Using seven cadaveric lumbosacral spines, the following three anterior reconstruction methods were sequentially performed at L4-5 level: anterior 3DF disc replacement; anterior BAK cages (BAK); and posterior pedicle screw fixation and anterior BAK cages combined (BAK + PS). The L2-3 level received two methods of posterior reconstructions: subtotal 3DF disc replacement (two implants), and posterior interbody cages and pedicle screw fixation (PLIF). Six unconstrained pure moments were applied and three-dimensional segmental motions were measured with an optoelectronic motion measurement system. The center of rotation (COR) calculation was conducted radiographically using flexion-extension films. Both anterior and posterior 3DF replacements statistically demonstrated equivalent range of motions (ROMs) in all loading modes compared to intact segment. Anterior BAK, BAK + PS, and PLIF demonstrated significantly lower ROMs when compared to intact and 3DF groups (P<0.05). The 3DF reconstruction tended to realign the COR to the posterior third or surrounding position at the operative disc level. The stand-alone lumbar 3DF disc replacement demonstrated biomechanical characteristics nearly equivalent to the intact spinal segments even through anterior or posterior approach in vitro, suggesting an excellent clinical potential.