To predict the viscosity of any given biodiesel fuel (FAME mixture), a novel topological index based on the distance matrix and adjacent matrix of the molecular structure is proposed. The new topological index can reflect the information of the molecular structure for fatty acid methyl ester (FAME), such as the size of molecular, unsaturated bond and branch degree. Combined with the modified Grunberg-Nissan or Hind equation, the topological index values of the FAME mixture were calculated. Then, relates the topological index values of the FAME mixtures with the viscosities of them, two linear regression equations were obtained. Using these regression equations, the viscosity of biodiesel fuels were predicted. The results show the modified Grunberg-Nissan equation with a higher precision of prediction than the Hind equation regression equation. (c) 2007 Published by Elsevier Ltd.