An 'artificial' plasma for one-stage factor-VIII assays is made by incubating human plasma with EDTA, to destroy factor VIII, and afterwards removing the anticoagulant by dialysis. Bovine factor V is then added to a given level. In the assay, contact activation is controlled by adding contact product. It was confirmed that factor-VIII activity was destroyed and that the EDTA was freely dialysable. The fibrinogen in the treated plasma clotted normally with thrombin. Likely variation in the factor-V activity was found not to be critical. The concentration of fibrinogen and other factors was adequate. Variation between batches was small. The artificial plasma yielded assay results closely comparable to haemophilic plasma in samples with factor-VIII activities in the range 0.01--20.0 iu/ml; the mean results in the artificial system were estimated to be 0.997 x those in haemophilic plasma, with a 95% confidence interval of 0.901--1.103. Biological variability in individual assays was smaller in the artificial system than when haemophilic plasma was used. Instability at the bench was more often detected in the artificial system than in haemophilic plasma assays, but the effect was eliminated from the results by obtaining duplicated readings in a balanced order.