This paper studies manual order picking activities in a warehouse where items are stored on pallets in two rows one above the other. Items stored on the pallets may be heavy, and manually picking the items may require excessive bending and stretching, which results in high spinal loads on the order picker associated with high injury risks. For this scenario that can frequently be observed in practice, this paper proposes an integrated model that supports the planning of order picking operations and pallet rotations taking account both of the time required for completing a set of orders as well as the spinal load on the order picker and the consequent injury risks. The results of a numerical experiment indicate that selectively rotating pallets may both reduce order picking time as well as the load on the order picker, leading to a quicker and less risky order picking process. The model proposed in this paper supports the decision of which pallet to rotate (and which not to rotate) against the company's cost objectives and its endeavor for worker well-being.