The hallmark of severe haemophilia, defined as a circulating level of factor (F) VIII (haemophilia A cases) or FIX (haemophilia B cases) of < 1%, is recurrent bleeding into muscles and joints (haemarthroses) from an early age of life. The inevitable result of such bleeding is progressive joint damage, leading to disabling arthritis that is typically evident within the first 2 decades of life in people with haemophilia who have limited or no access to regular factor replacement therapy, or those in whom factor replacement therapy is ineffective because of the presence of high-titre inhibitors. For children with severe haemophilia and no evidence of inhibitors, the unwanted musculoskeletal complications of severe haemophilia can be effectively prevented by the early initiation of a programme of long-term factor prophylaxis. In order to achieve the best outcome (a perfect musculoskeletal status for age) the programme of prophylaxis should be started before the onset of joint damage (primary prophylaxis). The gold standard primary prophylaxis regimen (the Malmo protocol) was pioneered and tested in Sweden and involves the infusion of 20-40 IU of FVIII per kg body weight on alternate days (minimum three times per week) for haemophilia A cases, and 20-40 IU kg(-1) of FIX twice weekly for haemophilia B cases. This protocol is, however, demanding on peripheral veins and very expensive. Modifications of the parent protocol such as starting primary prophylaxis with once-weekly infusions via peripheral veins with rapid escalation to full-dose prophylaxis or dose escalation based on frequency of bleeding are increasingly implemented in haemophilia treatment centres in countries that can afford the high cost of such programmes. These modified programmes can be achieved in the majority of young children with severe haemophilia without the need for central venous access devices (e.g. Port-a-Caths) and with avoidance of device-associated complications such as infection and thrombosis. In at least one centre, experience with arteriovenous fistulae as a strategy to ensure reliable venous access is being accumulated. The issues of compliance (adherence) to recommended prophylaxis protocols and when, if ever, to stop a programme of primary prophylaxis once started are real and require ongoing prospective studies. Such studies should incorporate outcome measures such as health-related quality-of-life and economic analyses.