Methods of modelling the female torso during physical activity often neglect the position and movement of the breast. This novel investigation compares three female torso modelling approaches that differ in complexity (integrated breast, fixed breast, dynamic breast) to determine the effect on spinal joint moments during running and jumping. The commonly used integrated breast model distributed breast mass within the torso, the fixed breast model attached the mass of the breasts to fixed positions on the anterior of the torso, and a new dynamic breast model enabled relative motion between the breasts and anterior torso. Key findings demonstrated minimal differences in lumbar spine moments (<0.05 Nm/kg; 4%) between integrated breast and fixed breast models but greater differences, up to 0.86 Nm/kg (68%) during running and 0.89 Nm/kg (82%) during jumping, when breast motion was included. Thoracic spine moments revealed similar patterns with minimal differences (<0.05 Nm/kg; 11%) between integrated breast and fixed breast models and greater differences, up to 0.48 Nm/kg (92%) during running and 0.63 Nm/kg (66%) during jumping, with the dynamic breast model. Future female musculoskeletal models should consider including breast mass and motion to avoid mis-representing spinal loading in females during running and jumping.