Coordinating knowledge transfer within multi-plant manufacturing networks is a challenging task. Using a computational model, we examine when it is beneficial to create production knowledge within a central unit, the “lead factory,” and transfer it to geographically dispersed plants. We demonstrate that the knowledge transfer generates a trade-off between a positive cost-saving effect due to fewer adaptations in each plant, and a negative transfer cost effect due to the costly knowledge transfer itself. The complexity of the production process moderates the performance implications of the knowledge transfer because it determines the relative strength of these two effects. For production processes with low complexity, knowledge transfer can engender superior network performance. Here, an optimal extent of knowledge transfer exists, and thus, a complete knowledge transfer is not performance maximizing. For production processes with medium and high levels of complexity, performance is reduced rather than enhanced through knowledge transfer so that it is optimal not to transfer any knowledge from the lead factory to the plants. While we analyze knowledge transfer within a manufacturing network, our results are transferable to other settings that consist of a knowledge sending and receiving unit.