In traditional machine scheduling, the focus is on determining the sequence of jobs on the machines. Meanwhile, the logistical question of how to organize the transport process between production stages frequently remains in the background. In this article, however, we consider lot streaming, offering the possibility to forward units to the next production stage before the entire production lot has been completed. Specifically, we analyze lot streaming in a hybrid flow shop where the aim is to minimize the makespan. For this purpose, two mixed-integer optimization models are developed and compared with respect to their solution quality. Based on the superior formulation, we analyze the influence of the configuration of the hybrid flow shop in terms of the number of production stages, the number of machines on the stages, and the length of the processing times on the makespan for moderate problem sizes. The results obtained highlight the importance and effect of different lot streaming scenarios in scheduling, and, in particular when the increased complexity of scheduling with sublots is worthwhile.