A recently introduced measure of bipartivity in complex networks is tested in the role of a selector of stable fullerene isomers. Its performance is compared with other proposed stability predictors on the set of experimentally characterized fullerene isomers. It is found out that the new predictor exhibits a high degree of selectivity. This, together with the simplicity of its implementation and low computational complexity, makes it a useful tool in narrowing the set of candidates for stable fullerene isomers. (c) 2005 Elsevier B.V. All rights reserved.