The management of unexpected bleeding must be directed at the specific abnormality identified, as there is no universally effective and safe procoagulant product. Where practical, a purely pharmaceutical approach obviates the residual risks of exposure to plasma-derived products. Desmopressin is often effective in bleeding due to mild haemophilia A, Type I von Willebrand's disease and some platelet function disorders. Where replacement therapy is necessary, it should be as specific as possible, preferably using purified components singly or in combination. Recombinant proteins provide the greatest margin of safety, but it must be borne in mind that these are biologicals, and that they may contain human and animal plasma-derived proteins. Where specific replacement is unavailable or impractical, plasma or crudely fractionated plasma derivatives may be used. In the case of inhibitor antibodies to factor VIII, high dose human factor VIII or porcine factor VIII may be used. Where replacement therapy is impossible due to a high inhibitory titre, it may be necessary to bypass the specific haemostatic defect using activated prothrombin complex concentrates or recombinant activated factor VIIa. The latter product is being studied in patients with various disorders of platelet function, and in the more global haemostatic failure that accompanies end-stage liver disease. Ancillary methods are often of great value in securing haemostasis. These may be derived from pharmacological or biological sources, and their sites of action may be systemic or topical. Examples include antifibrinolytic lysine analogues, corticosteroids where inflammation accompanies bleeding, and the topical application of fibrin sealants or thrombin. Simple physical measures such as pressure, ice, or splinting are also valuable adjunctive measures. Finally, it must be emphasized that the ultimate control of bleeding often depends upon effective management of the inciting cause, such as eliminating the trigger for DIC, or suppressing the causative antibody of ITP. These principles will be presented using a practical algorithmic approach. The initial question when considering treatment should be whether or not the patient is acutely unstable. Instability may be due to one of two causes: the volume of blood loss leading to a compromised cardio-vascular status, or the site of the bleed. The relevance of the site of the bleed is independent of the volume of blood loss, so for example, a closed bleed into CNS will cause critical functional compromise even though the volume of bleeding may be minimal. Similarly bleeding into a compartment, such as into a forearm or a calf will cause critical functional compromise irrespective of the volume of bleeding.