Previous analyses of developmentally programmed capillary regression suggested two distinct causes of vascular endothelial cell (VEC) death. The first appeared to be macrophage-dependent (Lang, R. A. and Bishop, M. J. (1993) Cell 74, 453-462) while the second was proposed to result from cessation of blood flow (Lang, R. A., Lustig, M., Francois, F., Sellinger, M. and Plesken, H. (1994). Development 120, 3395-3403). Combined, these analyses suggested a model in which initial, macrophage-mediated endothelial cell apoptosis blocked blood flow within a capillary segment and, as a consequence, caused apoptosis of all remaining cells in the affected segment. In the current study, we have tested this model using a new method that combines vital and histological analyses as a means of determining the fate of whole capillary segments and individual cells in vivo. This technique revealed that one of the first events in regression was the apoptosis of a single VEC in otherwise normal, flowing capillary segments (initiating apoptosis). These isolated, dying VECs projected into and restricted the capillary lumen, imposing either a temporary or permanent block to blood flow. Following cessation of flow, synchronous apoptosis of VECs occurred (secondary apoptosis). In addition, a quantitative analysis revealed a reciprocal relationship between plasma flow and VEC apoptosis. These observations are consistent with a model for capillary regression in which macrophages induce apoptosis in a limited number of VECs and, as a consequence of a block to blood flow, also cause apoptosis in those remaining.