Brain tumor dispersal far from bulk tumor contributes to and, in some instances, dominates disease progression. Three methods were used to characterize brain tumor cell motility in vivo and in vitro: 1) 2 weeks after implantation in rat cerebral cortex, single C6 cells labeled with a fluorescent tag had migrated to brain sites greater than 16 mm distant from bulk tumor; 2) time-lapse videomicroscopy of human brain tumor cells revealed motility of 12.5 microns/hr. Ruffling leading edges and pseudopod formation were most elaborate in more malignant cells; 3) an in vitro assay was devised to quantitatively evaluate motility from a region of high cell density to one of lower cell density. Human brain tumor cells were plated in the center of a petri dish, washed, and refed, establishing a 2-cm circular zone of cells in the dish center. Motility was determined by counting cells daily at predetermined distances from the central zone perimeter. Cells were found 1 cm from the perimeter by 24 hours and 3 cm from the perimeter by 4 days. Increasing serum concentration increased motility; however, neither fibronectin nor arrest of cells in the G0 phase by hydroxyurea altered motility. The addition of cytochalasin B to block cytoskeletal assembly prevented cell motility. Motility increased with increased malignancy. Subpopulations of cells were created by clonal amplification of cells that had migrated most rapidly to the dish periphery. Although morphologically indistinguishable when compared to the original cell line from which they were derived, these subpopulations demonstrated significantly increased motility.