Quantitative structure-activity relationship (QSAR) models are useful in providing a biochemical understanding of the biological activity of natural and synthetic chemicals based solely on molecular structure. One- to three-parameter multiregression equations were generated for three different groups of flavonoids to model experimental flavonoid-induced cytotoxicity and inhibition of human immunodeficiency virus I replication in lymphocyte-infected cells, using quantum chemical and geometrical parameters derived from optimized molecular structures. Both biological properties were basically dependent on electronic parameters describing charge distribution on the two fused rings of the flavonoid molecule. Atomic charges in C3 and the carbonyl carbon as well as the dipolar moment were important electronic descriptors to define the studied biological properties of flavonoids.