Abstract
Connectivity indices χ of the line graph L and higher line graphs L n , n = 2, 3,…, of the molecular graph G are examined as possible structure-descriptors in QSPR and QSAR studies. In the case of alkanes, regression models based on χ(G), χ(L), χ(L2), χ(L 3) and χ(L4) are found to be significantly better than those based solely on the index χ(G), with χ(G) continuing to play an important role. This has been demonstrated for boiling point, molar volume, molar refraction and surface tension. The same model was then applied for predicting the microsomal p-hydroxylation of aniline by alcohols. The QSAR model obtained by using connectivity indices of higher line graphs represents a significant improvement respect to the model using solely χ(G).