This study deals with designing and validating QSAR models generated for 45 flavonoids having PTP1B inhibition properties. Eight molecular descriptors/pharmacophoric features of each flavonoid along with their reported IC50 values against PTP1B were utilized to prepare training sets and generate models. It was developed by employing linear regression to calculate the predicted IC50 values. The generated models were validated using reported IC50 values of test sets. The correlation R2 values were observed to be in the following order, 92.45% (for an increasing hydrogen bond donor), 92.08% (for randomly sorting), 91.85% (for increasing molecular weight), 84.19% (for increasing hydrogen bond acceptor), 64.91% (for increasing TPSA), 53.90% (for increasing number of rotatable bonds) and 52.28% (for increasing log P); signifying the role of these pharmacophoric features while drug designing. Molecular docking of the flavonoids with the PTP1B active site revealed interactions with catalytic site and adjacent loops. The models would be beneficial for further studies for drug designing against PTP1B inhibition and therapeutic implications for treatment of insulin resistance.