The lower back is the most sensitive part of the human spine and over loading and bad posture during lifting can damage this area of the body. The lumbar spine consists of five vertebrae, which are responsible for carrying the weight of the upper body and loads. Intervertebral discs allow articulation between vertebrae. These discs are primarily made of non-homogeneous soft tissue, which allows the vertebrae to move and flex in all directions.Biomechanical models have been developed in the past decades to model and to predict the behavior of the spine in response to different loads. With the advances in computer modeling technology, analytical methods have become more popular in modeling the spine. These models are more cost effective and practical compared to the early models and use of human volunteers and cadavers. Unfortunately due to the complexity of the spine, most of the models failed to offer an accurate estimation of reaction moments and forces. Most models also use proprietary and custom-made software which makes it difficult for other researchers to use and modify them. This thesis reports the development and verification of a multi-body computational model of the lumbar spine. The model comprises five lumbar vertebrae (L1 to L5) and pelvis (S1). The vertebrae are connected to each other by invertebral discs, which consist of an anatomically correct kinematic and dynamic constraints. This combination represents a six degree-of-freedom mobility and enables the model to accommodate flexion, lateral bending, and axial rotation. The model is validated by carrying out a series of case studies including experimental motion studies. It is also used for preliminary evaluation of an ergonomical device called the dynamic trunk support (DTS), developed at Ryerson, School of Occupational and Public Health, in conjunction with the Mechanical and Industrial Engineering department. The results are in good agreement with the experimental results.