Cell polarity must be integrated with tissue polarity for proper development. The Drosophila embryonic central nervous system (CNS) is a highly polarized tissue; neuroblasts occupy the most apical layer of cells within the CNS, and lie just basal to the neural epithelium. Neuroblasts are the CNS progenitor cells and undergo multiple rounds of asymmetric cell division, ;budding off' smaller daughter cells (GMCs) from the side opposite the epithelium, thereby positioning neuronal/glial progeny towards the embryo interior. It is unknown whether this highly stereotypical orientation of neuroblast divisions is controlled by an intrinsic cue (e.g. cortical mark) or an extrinsic cue (e.g. cell-cell signal). Using live imaging and in vitro culture, we find that neuroblasts in contact with epithelial cells always ;bud off' GMCs in the same direction, opposite from the epithelia-neuroblast contact site, identical to what is observed in vivo. By contrast, isolated neuroblasts 'bud off' GMCs at random positions. Imaging of centrosome/spindle dynamics and cortical polarity shows that in neuroblasts contacting epithelial cells, centrosomes remained anchored and cortical polarity proteins localize at the same epithelia-neuroblast contact site over subsequent cell cycles. In isolated neuroblasts, centrosomes drifted between cell cycles and cortical polarity proteins showed a delay in polarization and random positioning. We conclude that embryonic neuroblasts require an extrinsic signal from the overlying epithelium to anchor the centrosome/centrosome pair at the site of epithelial-neuroblast contact and for proper temporal and spatial localization of cortical Par proteins. This ensures the proper coordination between neuroblast cell polarity and CNS tissue polarity.