Background: S2 cells are one of the most widely used Drosophila melanogaster cell lines. A series of studies has shown that they are particularly suitable for RNAi-based screens aimed at the dissection of cellular pathways, including those controlling cell shape and motility, cell metabolism, and host-pathogen interactions. In addition, RNAi in S2 cells has been successfully used to identify many new mitotic genes that are conserved in the higher eukaryotes, and for the analysis of several aspects of the mitotic process. However, no detailed and complete description of S2 cell mitosis at the ultrastructural level has been done. Here, we provide a detailed characterization of all phases of S2 cell mitosis visualized by transmission electron microscopy (TEM). Results: We analyzed by TEM a random sample of 144 cells undergoing mitosis, focusing on intracellular membrane and microtubule (MT) behaviors. This unbiased approach provided a comprehensive ultrastructural view of the dividing cells, and allowed us to discover that S2 cells exhibit a previously uncharacterized behavior of intracellular membranes, involving the formation of a quadruple nuclear membrane in early prometaphase and its disassembly during late prometaphase. After nuclear envelope disassembly, the mitotic apparatus becomes encased by a discontinuous network of endoplasmic reticulum membranes, which associate with mitochondria, presumably to prevent their diffusion into the spindle area. We also observed a peculiar metaphase spindle organization. We found that kinetochores with attached k-fibers are almost invariably associated with lateral MT bundles that can be either interpolar bundles or k-fibers connected to a different kinetochore. This spindle organization is likely to favor chromosome alignment at metaphase and subsequent segregation during anaphase. Conclusions: We discovered several previously unknown features of membrane and MT organization during S2 cell mitosis. The genetic determinants of these mitotic features can now be investigated, for instance by using an RNAi-based approach, which is particularly easy and efficient in S2 cells.