The blood coagulation disorder, hemophilia A, is caused by de®ciency of coagulation factor (F)VIII. Hemophilia A is now treated by infusions of pure FVIII, but the activity of FVIII is limited because it is unstable following activation by thrombin. This instability of activated FVIII is the result of dissociation of the A2 subunit. To obtain increased stability in FVIIIa, a disul®de bond between the A2 domain and the A3 domain, preventing A2 subunit dissociation, has been engineered. Structural analysis of the FVIII A domain homology model allowed us to identify residues 664 and 1826 as a potential disul®de bond pair. A FVIII mutant containing Cys664 and Cys1826 was produced and puri®ed (C664-C1826 FVIII). Immunoblotting showed that a disul®de bond did form to link covalently the A2 and the A3 domains. Following activation of the recombinant C664-C1826 FVIII by thrombin, the mutant FVIIIa had increased stability and retained more than 90% of its clotting activity at a time at which wild-type FVIIIa lost more than 90% of its activity. This remarkably stable C664-C1826 FVIIIa provides a unique approach for studies of the cofactor activity of FVIIIa and also for new, improved therapy for hemophilia A.