We examined the formation of motile, chemotactically active, anucleate fragments from human blood polymorphonuclear leukocytes (PMN, granulocytes), induced by the brief application of heat. These granule-poor fragments are former protopods (leading fronts, lamellipodia) that become uncoupled from the main body of the cell and leave it, at first with a connecting filament that breaks and seals itself. The usual random orientation of such filaments can be controlled by preorientation of cells in a gradient of the chemotactic peptide, N-formylmethionylleucylphenylalanine (F-Met-Leu-Phe) (2x10(-9) M- 1x10(-8)). Cytochalsin B, 2.5-5 mug/ml, prevents fragment formation; colchicine, 10(-5) M, does not. In scanning electron micrographs, fragments are ruffled and the cell body rounded up and rather smooth. In transmission electron micrographs, fragments contain microfilaments but lack centrioles and microtubules. Like intact cells, both bound and free fragments can respond chemotactically to an erythrocyte destroyed by laser microirradiation (necrotaxis); the free, anucleate fragments may do so repeatedly, even after having been held overnight at ambient temperatures. We propse the name cytokineplast for the result of this self-purification of motile apparatus. The exodus of the motile machinery from the granulocyte requires anchoring of the bulk of the cell to glass and uncoupling, which may involve heat-induced dysfunction of the centrosome. In ultrastructural studies of the centrosomal region after heat, centriolar structure remains intact, but pericentriolar osmiophilic material appears condensed, and microtubules are sparse. These changes are found in all three blood cell types examined: PMN, eosinophil, and monocyte. Of these, the first two make fragments under our conditions; the more sluggish monocyte does not. Uncoupling is further linked to centrosomal dysfunction by the observation that colchicines-treated granulocytes (10(-5)M, to destroy the centrosome's efferent arm) make fragments after less heat than controls. If motive force and orientation are specified mainly from the organelle-excluding leading front, then endoplasmic streaming in PMN is a catch-up phenomenon, and microtubules do not provide the vector of locomotion but rather stabilize and orient the "baggage" (nucleus, granuloplasm)-i.e., they prevent fishtailing. Moreover, constraints emanating from the centrosome may now be extended to include, maintenance of the motile machinery as an integral part of the cell.