A method for developing 3D synthetic capillary networks representative of the geometric, topological and functional properties of the cerebral microcirculation is presented. Previous methods [2, 4] have lacked quantitative comparison against anatomical cerebral networks. It is hypothesised that healthy capillary networks are space-filling, ensuring that the maximum separation of each tissue cell from the nearest vessel is less than a certain length, linked to the diffusion limited distance of oxygen transport in oxygen consuming tissue. Building on this theory, Lorthois & Cassot [3] constructed 2D synthetic space-filling networks by randomly placing one point in each cell of an n × n grid. From this set of points, representing the centres of tissue cells, Voronoi diagrams were extracted with the edges producing a 2D network of capillaries with three segments at each node, a characteristic feature of cerebral capillaries networks (see Figure 1, left). The goal of the present work is to extend this approach to 3D and to compare the resulting structural and hemodynamic properties to the properties of anatomical cerebral capillary networks.