Coatings from diamond-like carbon (DLC) have been proven to be an excellent choice for wear reduction in many technical applications. However, for successful adaption to the total joint replacement field, layer performance, stability and adhesion in realistic physiological setups are quintessential and these aspects have not been consistently researched. In our team’s efforts to develop long-term stable DLC implant coatings, test results gained from a simplified linear spinal simulator setup are presented. It is shown that metal-on-metal (MoM) pairs perform well up to 7 million loading cycles, after which they start to generate wear volumes in excess of 20 times those of DLC-coated implants. This is attributed to the roughening observed on unprotected metal surfaces. Furthermore, we illustrate that in contrast to DLC-on-DLC, MoM tribopairs require protein-containing media to establish low-friction conditions. Finally, results of defect monitoring during testing are presented, showing catastrophic failure of layers whose interfaces are too weak with respect to the stress-corrosion-cracking mechanism encountered in vivo.