An important production planning problem is how to best schedule jobs (or lots) when each job consists of a large number of identical parts. This problem is often approached by breaking each job/lot into sublots (termed lot streaming). When the total number of transfer sublots in lot streaming is large, the computational effort to calculate job completion time can be signif-icant. However, researchers have largely neglected this computation time issue. To provide a practical method for production scheduling for this situation, we propose a method to address the n-job, m-machine, and lot streaming flow-shop scheduling problem. We consider the variable sublot sizes, setup time, and the possibility that transfer sublot sizes may be bounded because of capacity constrained transportation activities. The proposed method has three stages: initial lot splitting, job sequencing optimi-zation with efficient calculation of the makespan/total flow time criterion, and transfer adjustment. Computational experiments are conducted to confirm the effectiveness of the three-stage method. The experiments reveal that relative to results reported on lot streaming problems for five standard datasets, the proposed method saves substantial computation time and provides better solu-tions, especially for large-size problems.