It has been well documented that there is an uncertainty over the true factor (F)VIII level in postinfusion samples due to assay discrepancies. The thrombin generation test (TGT) was used as a potentially more physiological approach to assess and compare FVIII concentrates. FVIII concentrates were added to artificial FVIII-deficient plasma. Thrombin generation was initiated by the addition of FIXa (14 nm), phospholipid and CaCl2. Thrombin was measured by subsampling into fibrinogen, and curves quantified as area under the curve (AUC) and time taken to half-maximum (t(1/2)max). Addition of one plasma-derived concentrate to as little as 0.005 IU mL-1 gave a normal AUC, but prolonged t(1/2)max. Increasing FVIII to 1 IU mL-1 had little effect on AUC, but did reduce the t(1/2)max to 64 s (normal 114 s). A range of plasma-derived and recombinant concentrates were tested at 1 IU mL-1; results were similar, except the B-domain deleted concentrate, which had the most rapid initial rate of thrombin generation (t(1/2)max 48 s, P < 0.05). Two hemophilic plasmas (< 0.01 IU mL-1) produced large amounts of thrombin (AUC 65% and 69%), although t(1/2)max was prolonged. Addition of a FVIII antibody abolished thrombin generation, indicating that these plasmas contained low levels of FVIII. Decreasing the FIXa concentration (0.2 nm) minimized thrombin generation in hemophilic plasma but not in normal plasma. These results indicate that FVIII < 0.01 IU mL-1 can generate significant quantities of thrombin depending upon the amount of FIXa present. The TGT could prove useful for patient monitoring in gene therapy and prophylaxis.