Kallistatin is a serpin first identified as a specific inhibitor of tissue kallikrein. Our recent studies showed that kallikrein promoted angiogenesis, whereas kallistatin inhibited angiogenesis and tumor growth. This study is aimed to identify the structural elements of kallistatin essential for its antiangiogenic function. Kallistatin mutants at the hinge region (A377T) and a major heparin-binding domain (K312A/K313A) were created by site-directed mutagenesis. Recombinant kallistatin mutant A377T did not bind or inhibit tissue kallikrein activity. Wild-type kallistatin and kallistatin mutant A377T, but not kallistatin mutant K312A/K313A lacking heparin-binding activity, inhibited VEGF-induced proliferation, growth, and migration of human microvascular endothelial cells. Similarly, wild-type kallistatin and kallistatin mutant A337T, but not kallistatin mutant K312A/K313A, significantly inhibited VEGF-induced capillary tube formation of cultured endothelial cells in Matrigel and capillary formation in Matrigel implants in mice. To elucidate the role of the heparin-binding domain in modulating angiogenesis, we showed that wild-type kallistatin interrupted the binding of ¹²⁵I-labeled VEGF to endothelial cells, whereas kallistatin mutant K312A/K313A did not interfere with VEGF binding. Consequently, wild-type kallistatin, but not kallistatin mutant K312A/K313A, suppressed VEGF-induced phosphorylation of Akt. Taken together, these results indicate that the heparin-binding domain, but not the reactive site loop of kallistatin, is essential for inhibiting VEGF-induced angiogenesis.