The unique marine ecosystem is more and more recognized as a source of extremely potent natural anticancer agents. The Molecular Electronegativity Distance Vector (MEDV-13) was used to describe the chemical structure of 44 Fascaplysin analogues, and resulting in a quantitative structureproperty relationship (QSAR) model of six parameters on IC₅₀ for CDk4 using variable selection and modeling based on prediction (VSMP), which was validated ty LOO method and external test set were respectively. The obtained model shows good estimation ability and strong predictive power for the external samples with a calibrated correlation coefficient of r=0.923 9 and LOO validated correlation coefficient of q=0.878 9. Last, the influences of each index on activity were calculated by its relative or fraction contribution. The results revealed that the main structural factors influencing the bioactivities are =C-(or-C-)、>C-、>N-、-O and benzene site in biphenyl, and parabiphenyl is advantageous to activity.