Objective. To estimate the three-dimensional spinal loads during various lifting and lowering tasks. Design. The in vivo measurements of the trunk dynamics, moments, and myoelectric activity were used as inputs into an electromyographic-assisted model used to predict the three-dimensional spinal loads. Background. Previous studies of eccentric motions have investigated muscle activity, trunk strength, and trunk moments. A void in the body of knowledge exists in that none of these studies investigated spinal loading. Methods. Ten subjects lifted (40 ° of flexion to 0 °) and lowered (0 ° of flexion to 40 °) boxes while positioned in a structure that restrained the pelvis and hips. The tasks were performed under isokinetic trunk velocities of 5, 10, 20, 40, and 80 deg s −1 while holding a box with weights of 9.1, 18.2, and 27.3 kg. Results. Lowering strength was found to be 56% greater than lifting strength. The lowering tasks produced significantly higher compression forces but lower anterior-posterior shear forces than the lifting tasks. The differences in the spinal loads produced by the two lifting tasks were attributed to differences in coactivity and unequal lifting moments (i.e. holding the box farther away from the body). Conclusions. The nature of the spinal loads that occur during lowering and lifting were significantly different. The difference in spinal loads may be explained by different lifting styles.