The mechanism of the known Stone-Wales rearrangement of bifluorenylidene to dibenzo[g,p]chrysene is assessed with the aid of B3LYP/6-31G(d) density functional calculations, and it is shown that a radical-promoted mechanism involving a sequence of homoallyl-cyclopropylcarbinyl rearrangement steps gives a realistic activation energy and can explain experimental observations, whereas a unimolecular mechanism has an improbably high activation energy. Radical-promoted mechanisms are then applied to the hypothetical Stone-Wales rearrangements of diindeno[1,2,3,4-defg;1',2',3',4'-mnop]chrysene and C-60 itself. Severe steric constraints in these cases raise the activation energy for the radical-promoted pathways substantially, but they are still strongly preferred to uncatalyzed, unimolecular pathways.