Pathological blood clotting (thrombosis) is a leading cause of cardiovascular disease, which causes an estimated 17.9 million deaths globally each year. Existing drugs for thrombosis mainly target proteases from the coagulation system, but they carry side effects, e.g. the risk of excessive bleeding, because they inhibit proteases that are also required for blood clotting after injury. Recently, coagulation factor XIIa (FXIIa) emerged as a promising target because it has a key role in disease-related blood clotting but not for injury-related blood clotting. Previously, a cyclic peptide inhibitor of FXIIa was designed by altering several contact residues in the cyclotide, Momordica cochinchinensis trypsin inhibitor-II (MCoTI-II). However, this inhibitor has limited potency and lacks selectivity over key off-targets. Our study considered an alternative approach and focused on structural optimisation in loop 6 of MCoTI-II. MCoTI-based inhibitors with different numbers of residues in loop 6 were produced to identify new loop configurations that improved the inhibitor’s activity. Residues in loop 6 of the best variant were scanned by Ala substitution to examine the importance of each residue for activity. A key contact residue was subsequently optimised by substituting FXIIa preferred residues identified in an additional peptide library screen. Overall, a potent MCoTI-based inhibitor was engineered for FXIIa (K_i = 2 nM) with 100-fold improved activity, with the new cyclic peptide representing a potential drug lead for the treatment of thrombosis.