Theoretically, von Willebrand factor (VWF) should be capable of binding all factor VIII (FVIII), but an unbound FVIII (uFVIII) plasma fraction remains. In patients' status post deep-vein thrombosis (DVT), an altered uFVIII fraction and high FVIII levels might be indicative of dysfunctional FVIII regulation. Out of 928 consecutive DVT patients, 321 were found to have high FVIII levels. After excluding 183 patients with known causes for high FVIII levels, plasma samples with unexplainably high FVIII levels were available from 84 patients. To capture the FVIII-VWF-complex, superparamagnetic polystyrene beads with covalently attached streptavidin were coated with biotinylated anti-rabbit Ig and incubated with rabbit anti-human VWF-Ig. Slowly thawed plasma samples were added to cooled beads, which were then separated by a magnetic particle concentrator. The uFVIII fraction was calculated by dividing the FVIII activity in the supernatant of the FVIII-VWF-complex-free sample by the FVIII activity in the supernatant of the control sample. Additionally, the VWF residuum in the supernatant was determined. Compared to age- and sex-matched blood donors, thrombosis patients showed a significantly higher plasma FVIII/VWF ratio (median: 1.3 vs. 1.0, p<0.001). uFVIII fraction data were adjusted for VWF residuum. After forward stepwise logistic regression, uFVIII had an odds ratio of 0.48 (95% CI 0.34-0.65), i.e. the uFVIII fraction was reduced in thrombosis patients. Analysis of covariance confirmed these results: In thrombosis patients, the estimated mean of the uFVIII fraction was significantly lower (6.34% vs. 7.58%, p<0.001). In conclusion, thrombosis patients with high FVIII levels showed a higher FVIII/VWF ratio, similar to mice with defective FVIII clearance. The clearly reduced uFVIII fraction lends further support to the hypothesis of a modified FVIII clearance.