Comparative molecular field analysis (CoMFA) and comparative molecular similarity indices analysis (CoMSIA) were applied to the ACE-inhibitory, antimicrobial, and bitter-tasting peptides. Predictive 3D-QSAR models were established using SYBYL multifit molecular alignment rule over a training set and a test set. The optimum models were all statistically significant with cross-validated coefficients (Q(2)) >0.5 and conventional coefficients (R(2)) >0.9, indicating that they were reliable enough for activity prediction. The obtained results may aid in the design of novel bioactive peptides.