Interfacing neurons with silicon semiconductors is a challenge being tackled through varied bioengineering approaches. The resulting constructs inform our understanding of neuronal coding and learning, and ultimately guide us towards creating intelligent neuroprostheses and brain-machine interfaces. A fundamental pre-requisite is to dictate the spatial organization of neuronal cells on silicon. We fabricated photolithographically defined designs of parylene-C on SiO2 wafers. These chips are activated with fetal calf serum, resulting in a culture environment that is differentially cell-adhesive (parylene) or repulsive (SiO2). LUHMES (Lund human mesencephalic) neurons cultured in isolation find the entire chip cell-repulsive. Applying another cell type (HEK 293 cells) enables LUHMES neurons to secondarily adhere. Over 3-4 days, patterned neurons extend neurites and form networks. Seeking to gain better control of neurite behaviour, we created different geometric arrays of parylene. The basic pattern was a circular node of parylene with four ‘spokes’ at 0°, 90°,180°, and 270°. Three different versions were trialed (node dia. 50μm, spoke length 125μm; node dia. 100μm, spoke 100 μm; node dia. 250μm, spoke 100 μm). Nodes were arranged orthogonally with respect to one another, with a 100μm interval separating adjacent node/spoke complexes. To quantify the impact of the parylene/SiO2 construct on neurite orientation, neurites were traced from start point (centred on a parylene node) to end point (defined either as branching point, termination, or encountering another cell body). Traced segments were divided into 100μm sub-segments. A tangent was taken to each sub-segment and the angle, θ, of each segment measured and categorized into 11.25° intervals. This process was conducted for areas encompassing each of the three different node geometries, and for differentiated LUHMES cultured randomly on a polystyrene surface treated to promote homogenous cell adhesion. Entries 180° apart were summed. 12 independent chip experiments enabled acquisition and measurement of 200 neurite sub-segments (each of 100μm length). A radial plot (figure 1) illustrates the orientation of neurite segments. Neurite directionality differed significantly according to parylene geometry (Kolmogorov-Smirnov tests: unpatterned substrate vs 250μm diameter nodes D=0.200, P=0.001; 250μm vs 100μm nodes D=0.204, P=0.001; 100μm vs 50μm nodes D=0.147, P=0.028). There is a trend towards increasingly orthogonal growth as parylene configuration changes from 250μm to 100μm to 50μm diameter nodes. For the generation of patterned neuronal networks capable of meaningful interrogation, there must be control over neurite growth direction and connectivity. We have developed a method of generating orthogonally arranged neuronal networks.