Examples of gold and silver anisotropic colloids, such as prisms and rods, have appeared in the literature for many years. In most cases, the morphologies of these thermodynamically unfavorable particles have been explained by the particular reaction environment in which they were synthesized. The mechanisms used to explain the growth generally fall into two categories, one in which chemically adsorbed molecules regulate the growth of specific crystal faces kinetically, and the other where micelle-forming surfactants physically direct the shape of the particle. This paper raises questions about the growth of anisotropic metal colloids that the current mechanisms cannot adequately address, specifically, the formation of multiple shapes in a single homogeneous reaction and the appearance of similar structures in very different synthesis schemes. These observations suggest that any growth mechanism should primarily take into consideration nucleation and kinetics, and not only thermodynamics or physical constrictions. The authors suggest an alternative mechanism where the presence and orientation of twin planes in these face-centered cubic (fcc) metals direct the shape of the growing particles. This explanation follows that used for silver halide crystals, and has the advantage of explaining particle growth in many synthesis methods. In this mechanism, twin planes generate reentrant grooves, favorable sites for the attachment of adatoms. Shape and structural data are presented for gold and silver particles synthesized using several different techniques to support this new model. Triangular prisms are suggested to contain a single twin plane which directs that growth of the initial seed in two dimensions, but limits the final size of the prism. Hexagonal platelets are suggested to contain two parallel twin planes that allow the fast growing edges to regenerate one another, allowing large sizes and aspect ratios to form. Rods and wires were found to have a fivefold symmetry, which may only allow growth in one dimension. It is expected that a superior mechanistic understanding will permit shape-selective synthesis schemes to be developed.