Following proteolytic activation of factor VIII (FVIII) by thrombin, the FVIIIa A2 domain, A3 domain and light chain (A3-C1-C2 domains) form a non-covalent hetero-trimer. Because spontaneous A2 subunit dissociation causes loss of FVIIIa activity, we previously made two mutants each with two new Cys to form a disulfide bond linking residues 662 (A2) and 1828 (A3) (FVIIIC662-C1828) or residues 664 (A2) and 1826 (A3) (FVIIIC664-C1826). Following thrombin activation, each FVIIIa mutant was stabile compared to wild type (wt) B-domain-deleted (BDD) FVIII. Previous SDS-PAGE data showed that the A2 domain was disulfide linked to the light chain. To show that this is true for undenatured FVIIIa, here we used surface plasmon resonance (SPR) to monitor A2 dissociation from thrombin-activated wild type and variant FVIII species that were bound to the sensor surface via a monoclonal antibody. Following passage of thrombin over sensor-bound FVIII, only wt FVIII showed a characteristic decrease of SPR reflecting A2 subunit dissociation and thrombin-treated FVIIIC662-C1828 and FVIIIC664-C1826 showed only minor decreases in SPR. Thus, SPR data directly demonstrate that engineered inter-domain disulfide bridges between the A2 and A3 domains prevent A2 domain dissociation from FVIIIa. In contrast to simple plasma coagulation assays of FVIIIa, rotational thromboelastogram (RoTEG) assays of whole blood provide multiple parameters reflecting clot formation, clot quality, and clot dissolution. RoTEG assays using fresh severe hemophilia A whole blood that was reconstituted with either wt FVIII, or FVIIIC662-C1828 or FVIIIC664-C1826 were performed to test the hypothesis that the disulfide-stabilized FVIIIa mutants would show improved potency for thrombin generation. After recalcification of hemophilia A blood with added FVIII, we measured the clotting time (CT), the rate of clot-formation, the clot-firmness time (CFT), defined as the time required to reach a specified clot firmness, and the clot firmness at 5 min (CF-A5), defined as the clot firmness at 5 min after the observed CT. Samples reconstituted with disulfide-bridge-stabilized FVIII mutants or wt-FVIII had comparable CTs at similar concentrations. However, in comparison to wild type BDD-FVIII, comparable rates of clot-formation, CFTs and CF-A5 were observed for up to 10-fold lower concentrations of each disulfide-bridge-stabilized FVIII mutant. The differences between wt and FVIII mutants were especially pronounced at very low FVIII concentrations whereas at FVIII concentrations >0.01 U/mL the differences were less apparent. Because clot formation occurs early relative to overall thrombin generation which is better reflected by CFT and CF-A5 values, we interpret these data to indicate that the disulfide-stabilized FVIIIa variants provide sustained thrombin generation in whole blood compared to wt FVIII and speculate that these FVIII variants may prove superior to wt FVIII for stabilizing a hemostatic plug by providing sustained thrombin generation capacity.