The purpose of this paper is to outline a technique for simulating the evolution of plants. The problem is addressed using genetic algorithms (GA's) [3, 5] operating on L-systems [12], with fractal dimension of the resulting structures serving as the fitness criterion. In physical terms, fractal dimension correlates with the "bushiness" or "fullness" of a plant. A plant with low fractal dimension (close to one) may have insufficient surface area to capture ample amounts of light. On the other hand, a plant with high fractal dimension (close to two in the case of planar structures) may be excessively bushy, causing substantial self-shading. Ultimately, plants which are bushy to the extent that they shade other plants but not themselves should be given evolutionary preference.