Force arising from myosin activity drives a number of different types of motility in eukaryotic cells. Outside of muscle tissue, the precise mechanism of myosin-based cell motility is for the most part theoretical. A large part of the problem is that, aside from cell surface features such as lamellipodia and microvilli, relatively little is known about the structural organization of potential actin substrates for myosin in non-muscle motile cells. Several groups [Cramer, Siebert and Mitchison (1997) J. Cell Biol. 136, 1287-1305; Guild, Connelly, Shaw and Tilney (1997) J. Cell Biol. 138, 783-797; Svitkina, Verkhovsky, McQuade and Borisy (1997) J. Cell Biol. 139, 397-415] have begun to address this issue by determining actin organization throughout entire non-muscle motile cells. These studies reveal that a single motile cell comprises up to four distinct structural groups of actin organization, distinguished by differences in actin filament polarity: alternating, uniform, mixed or graded. The relative abundance and spatial location in cells of a particular actin organization varies with cell type. The existence in non-muscle motile cells of alternating-polarity actin filament bundles, the organization of muscle sarcomeres, provides direct structural evidence that some forms of motility in non-muscle cells are based on sarcomeric contraction, a recurring theory in the literature since the early days of muscle research. In this scenario, as in muscle sarcomeres, myosin generates isometric force, which is ideally suited to driving symmetrical types of motility, e.g. healing of circular wounds in coherent groups of cells. In contrast, uniform-polarity actin filament bundles and oriented meshworks in cells allow oriented movement of myosin, potentially over relatively long distances. In this simple 'transport-based' scenario, the direction in which myosin generates force is inherently polarized, and is well placed for driving asymmetrical or polarized types of motility, e.g. as expected for long-range transport of membrane organelles. In the more complex situation of cell locomotion, the predominant actin organization detected in locomoting fish keratocytes and locomoting primary heart fibroblasts excludes sarcomeric contraction force from having a major role in pulling these cell types forward during locomotion. Instead Svitkina et al. propose that 'dynamic network contraction' of a weakly adherent uniform-polarity actin filament meshwork is the basis of keratocyte locomotion. For fibroblast locomotion, however, Cramer et al. prefer a transport mechanism based on graded-polarity actin filament bundles.