Abstract: Box loading is a critical intermediate stage in the order picking and delivery process. Box dimensions and stacking configurations have a significant impact on overall logistics service efficiency; however, this aspect is often overlooked in optimization studies on order picking and delivery. This study focuses on the joint scheduling problem of order picking and delivery. Under delivery sequence constraints, it incorporates a three-dimensional bin-packing problem with weakly heterogeneous boxes and develops an integrated picking–loading–delivery scheduling model (IOPDS-WHB). Furthermore, an improved simulated annealing algorithm (ISA) based on a two-stage framework is proposed to determine the box stacking configuration and picking scheme, with the objective of minimizing the total order fulfillment time. The results show that the ISA algorithm can effectively solve the IOPDS-WHB problem. Moreover, the proposed model improves overall logistics service efficiency, outperforming both the commonly used reverse picking strategy based on delivery sequence and the shortest picking-time strategy in existing studies that ignore the loading stage. Specifically, total operation time is reduced by 24.5% and 13.9%, respectively. These results demonstrate the effectiveness and practical applicability of the proposed IOPDS-WHB framework, which provides valuable managerial insights for optimizing order picking and delivery in real-world e-commerce logistics systems.