Finite Element models of the lumbar spine are commonly used for the study of spine mechanics and device performance, but have limited usefulness in some applications such as clinical and design phase assessments due to long analysis times. In this study a computationally efficient L4-L5 FSU model and a L1-Sacrum multi-segment model were developed and validated. The FSU is a functional spine unit consisting of two adjacent vertebral bodies, in this case L4 and L5. The multi-segment model consists of all lumbar vertebrae and the sacrum. The models are able to accurately predict spine kinematics with significantly reduced analysis times, relative to fully deformable representations. Analysis times were reduced from 3 hrs and 20 min to 2 min and 1 min for the multi-segment and FSU models, respectively. The vertebrae geometries were reconstructed from CT scans of the cadaveric specimen. Prior to model development, experimental testing was performed on the specimen using a custom multi-axis spine simulator. Collection of kinematic data in response to external loading made tuning of the model stiffness possible. The improved computational efficiency of the models makes them more useful for applications requiring multiple iterations and short analysis times such as clinical and design phase assessments of implants. The model can also be used in efforts to develop lumbar musculoskeletal models, which may require multiple runs for the optimization of muscle forces.