I consider a brush consisting of dendritic polymer end-grafted by its initial segment to a flat, impenetrable surface, in both the molten and solvated cases. The parabolic profile of linear grafted brushes is maintained, but the distribution of the free tips is quite different from the linear brush case. In particular, free ends are comparatively buried in the brush or "forest" layer compared to linear chains, an effect which is magnified by going to higher and higher generations. There is no appreciable tendency for dendritic arms to fold backward into the layer, however. The results of the classical path analysis compare favorably to lattice self-consistent-field calculations on the same systems.