We recently found that integrin αvβ3 binds to fibroblast growth factor (FGF)-αvβ31 (FGF1), and that the integrin-binding defective FGF1 mutant (Arg-50 to glutamic acid, R50E) is defective in signalling and antagonistic to FGF1 signalling. R50E suppressed angiogenesis and tumour growth, suggesting that R50E has potential as a therapeutic. However, FGF1 is unstable, and we had to express R50E in cancer cells for xenograft study, since injected R50E may rapidly disappear from circulation. We studied if we can develop antagonist of more stable FGF2. FGF2 is widely involved in important biological processes such as stem cell proliferation and angiogenesis. Previous studies found that FGF2 bound to αvβ3 and antagonists to αvβ3 suppressed FGF2-induced angiogenesis. However, it is unclear how FGF2 interacts with integrins. Here, we describe that substituting Lys-119/Arg-120 and Lys-125 residues in the predicted integrin-binding interface of FGF2 to glutamic acid (the K119E/R120E and K125E mutations) effectively reduced integrin binding to FGF2. These FGF2 mutants were defective in signalling functions (ERK1/2 activation and DNA synthesis) in NIH3T3 cells. Notably they suppressed, FGF2 signalling induced by WT FGF2 in endothelial cells, suggesting that the FGF2 mutants are antagonists. The FGF2 mutants effectively suppressed tube formation in vitro, sprouting in aorta ring assays ex vivo and angiogenesis in vivo The positions of amino acids critical for integrin binding are different between FGF1 and FGF2, suggesting that they do not interact with integrins in the same manner. The newly developed FGF2 mutants have potential as anti-angiogenic agents and useful tools for studying the role of integrins in FGF2 signalling.