Hemostasis is a complex physiological process responsible for the prevention of blood loss caused by vascular injury. The dysregulation of this delicate mechanism can lead to thromboembolic diseases, the leading cause of death worldwide. Anticoagulants are drugs applied for the treatment and prevention of this class of diseases. Their major side effect - severe bleeding, which is a life-threatening condition - strongly limits their potential and leads to events of inappropriate treatment. Thus, there is a strong unmet need for a safer class of anticoagulants. Coagulation factor XI (FXI), a serine protease from the intrinsic pathway of coagulation, has been identified as a promising target for safer anticoagulation. Studies in various animal models and epidemiological data have shown that FXI is involved in the pathogenesis of thrombosis, while not being essential for hemostasis. Various FXI-targeting molecules are currently in development, including small molecules, monoclonal antibodies, and antisense oligonucleotides. Peptide-based FXIa inhibitors have not yet been developed, despite the attractive properties of peptides as a drug format, including high affinity and selectivity, tunable pharmacokinetic properties, a fast onset of action and absence of toxic metabolic products. The aim of my thesis was the development of cyclic peptide-based FXIa inhibitors for safer anticoagulation. In a first project, I have developed a large cyclic peptide phage display library with unprecedented structural diversity. In a proof-of-concept selection performed with the library, peptides displaying long consensus sequences were enriched and potent binders were isolated, confirming the importance of having access to large and structurally diverse libraries. Next, I had applied the new phage display library for selecting cyclic peptide inhibitors of FXIa. Isolated peptides were characterized, and a particularly potent inhibitor was further optimized. The resulting peptide showed a strong anticoagulation effect ex vivo, which was comparable to the effect of heparin at therapeutic concentrations. I then modified the peptide, which improved its plasma stability and prolonged the circulation time in vivo. The modified peptide inhibited plasma coagulation after administration to rabbits and in an ex vivo model of hemodialysis using human blood. Finally, I had stumbled over a cyclic peptide FXIa inhibitor that had a moderate inhibitory constant but a particularly strong anticoagulation activity. In a structure-activity relationship analysis of the peptide, I identified the most important residue for the strong anticoagulation activity, which pointed to an electrostatic interaction and to potentially fast binding kinetics. The peptide indeed bound much faster than other FXIa inhibitors, showing that rapid engagement with the FXIa is important for efficient anticoagulation. Based on these findings, I concluded that it is worth to take into consideration the binding on-rate of inhibitors of coagulation proteases in the future development of anticoagulants.