Lumbar disc arthroplasty is a surgical procedure designed to treat degenerative disc disease by replacing the affected disc with a mobile prosthesis. Several types of implants fall under the term total disc replacement, such as ball-and-socket, mobile core or elastic prostheses. Some studies have shown that facet arthritis can develop after arthroplasty, without much precision on the mechanical impact of the different implant technologies on the facet joints. This study aims to create validated patient-specific finite element models of the intact and post-arthroplasty lumbar spine in order to compare the mechanical response of ball-and-socket and elastic prostheses. Intact models were developed from CT-scans of human lumbar spine specimens (L4-S1), and arthroplasty models were obtained by replacing the L4-L5 disc with total disc replacement implants. Pure moments were applied to reproduce physiological loadings of flexion/extension, lateral bending and axial rotation. Models with ball-and-socket prosthesis showed increased values in both range of motion and pressure at the index level and lower values at the adjacent level. The mechanical behaviour of the elastic prosthesis and intact models were comparable. The dissipated friction energy in the facet joints followed a similar trend. Although both implants responded to the total disc replacement designation, the mechanical effects in terms of range of motion and facet joint loads varied significantly not only between prostheses but also between specimens. This confirms the interest that patient-specific surgical planning using finite element analysis could have in helping surgeons to choose the appropriate implant for each patient.