23 obligate carriers of mild and severe haemophilia A and 26 normal females were bled on three occasions, and their plasmas assayed for procoagulant factor VIII (VIIIC), factor VIII related antigen (VIIIRAg) and factor VIII clotting antigen (VIIICAg). A comparison of the ratios VIIIC/VIIIRAg and VIIICAg/VIIIRAg indicated that, although the two ratios gave the same proportional misclassification of carriers as normals (four of 23), the latter ratio showed greater discriminatory power when an unequal variances predictive method was used to calculate likelihood ratios (for carrier status). This greater power was shown to be due to a greater reproducibility between visits for the VIIICAg/VIIIRAg ratio. Discrimination was considerably better when the median of the three median values for each variable was analysed, compared to the median value obtained at the first visit. There was also no statistical difference between VIIICAg/VIIIRAg (or VIIIC/VIIIRAg) ratios obtained from carriers of severe compared to mild haemophilia.