Total disc arthroplasty (TDA) has been successfully used for monosegmental treatment in the last few years. However, multi-level TDA led to controversial clinical results. We hypothesise that: (1) the more artificial discs are implanted, the stronger the increases in spinal mobility and facet joint forces in flexion and extension; (2) deviations from the optimal implant position lead to strong instabilities. A three-dimensional finite element model of the intact L1–L5 human lumbar spine was created. Additionally, models of the L1–L5 region implanted with multiple Charité discs ranging from two to four levels were created. The models took into account the possible misalignments in the antero-posterior direction of the artificial discs. All these models were exposed to an axial compression preload of 500 N and pure moments of 7.5 Nm in flexion and extension. For central implant positions and the loading case extension, a motion increase of 51% for two implants up to 91% for four implants and a facet force increase of 24% for two implants up to 38% for four implants compared to the intact spine were calculated. In flexion, a motion decrease of 5% for two implants up to 8% for four implants was predicted. Posteriorly placed implants led to a better representation of the intact spine motion. However, lift-off phenomena between the core and the implant endplates were observed in some extension simulations in which the artificial discs were anteriorly or posteriorly implanted. The more artificial discs are implanted, the stronger the motion increase in flexion and extension was predicted with respect to the intact condition. Deviations from the optimal implant position lead to unfavourable kinematics, to high facet joint forces and even to lift-off phenomena. Therefore, multilevel TDA should, if at all, only be performed in appropriate patients with good muscular conditions and by surgeons who can ensure optimal implant positions.