<p>Accurate estimation of spinal joint and muscle forces is important for understanding spine biomechanics and informing rehabilitation and surgical planning. Conventional upper-body musculoskeletal (MSK) models often lack thoracic spine detail and use generalized mass distributions. This study developed a physiologically detailed thoracolumbar MSK model with a flexible thorax (FT), articulated ribcage, and thoracic intervertebral disc (IVD) stiffness profiles to improve force prediction during 30° lateral bending. Nineteen male subjects (10 normal-weight, 9 obese) were simulated using constant percentage-based (CPB) and body shape-based (BSB) mass scaling. An indirect comparison with IDP-derived reference compressive-force estimates was used to assess the physiological plausibility of FT model predictions. During left lateral bending, BSB scaling reduced RMSE from 11.35 to 6.91 % BW in normal-weight subjects and from 41.05 to 16.98 % BW in obese subjects, whereas during right lateral bending RMSE remained nearly unchanged in normal-weight subjects and increased in obese subjects. After Holm-Bonferroni correction, statistically significant differences (p &lt; 0.05) between CPB and BSB models were observed in several joint and muscle force comparisons, particularly in obese subjects. Sensitivity analysis showed that compressive forces were highly dependent on thoracic disc stiffness, with interquartile range reaching 26.17 % BW at T12L1, whereas mediolateral shear forces varied by less than 2 % BW. These findings indicate that body-shape-based mass scaling and thoracic IVD rotational stiffness can substantially influence MSK-predicted spinal loads, particularly in obese subjects, and should be considered carefully when developing personalized thoracolumbar models.</p>