With the introduction of Level 3 Autonomous Driving Systems, an ability to maintain the appropriate level of protection for reclined occupants will become highly relevant. Virtual tools, such as human body models offer an attractive tool to support the rapid shift in vehicle development and restraint system design. However, previous research has shown that, due to differences in lumbar spine and pelvis flesh formulations, the available human body models, while valid in upright scenarios, differ in their responses in recline postures. The goal of this study is to use an updated lumbar spine and different pelvis flesh formulations to evaluate how these changes affect the response of reclined occupants in frontal crashes. An updated simplified lumbar spine was developed based on recently released GHBMC update. The updated simplified lumbar spine model was imported into previously utilised GHBMC simplified and detailed models, and an alternative pelvis flesh attachment was developed creating a total of four modified human body models. These models were evaluated in the sled environment representative of recent post-mortem human subject recline test series. The pelvis-flesh attachment model had a substantial effect on the submarining outcome across all evaluated models. With the unified updated simplified lumbar spine model both human body models showed differences in kinematic response and recorded lumbar forces.