The stress fiber (bundles of actin filaments) is one of the most prominent cytoskeletal components that contributes to the maintenance of cell architecture. It has generally been believed that upon cyclic stretching, both cells and their stress fibers become perpendicularly aligned to the direction of stretching. However, using our newly developed stretching device, we have recently found the contrary evidence that stress fibers in endothelial cells rapidly become rearranged at a specific oblique angle relative to the direction of stretching [Takemasa, T., K. Sugimoto, K. Yamashita: Exp. Cell Res. 230, 407-410 (1997)]. In light of this finding, we attempted to establish the explanation for such a phenomenon. First, we investigated the effects of possible modulators on the angle of the stress fibers; those were, modification of the stretching program, dependency of extracellular matrix types, and their reproducibility in other cell species. However, it seemed that the orientation was solely depending on the stretching amplitude applied. Next, we analyzed alterations in stress fiber length during loading tests using two kinds of deforming experiment systems. It was thus revealed that stress fibers aligned at a particular angle so as to minimize their length alterations in cyclic deforming fields. Rearrangement of the stress fibers at this angle probably occurs as a result of avoiding compressive stress and may be interpreted as a way of reducing the mechanical stress to which they are subjected during the deformation. This hypothesis well explains the reason not only for the survival of the stress fibers at a particular oblique angle, but also for the reduced numbers of stress fibers found at the other angles on cyclic deforming fields.