In order to improve our understanding of heparin bioavailability and clearance, and in an attempt to identify the factors responsible for variation in dose response, a reproducible in vivo model of heparin pharmacokinetics and pharmacodynamics was established in the rabbit. This has demonstrated that impaired heparin recovery is caused by increased non-specific protein binding. The extent of reversible protein binding is substantially less with low molecular weight than unfractionated heparin, and in contrast to unfractionated heparin which showed a significant reduction in recovery in experimental endotoxaemia, low molecular weight heparin does not bind to proteins that are released in response to endotoxin as its recovery was unaffected by acute endotoxaemia. The findings suggest that the binding proteins are likely to be proteins released from activated platelets, endothelial cells or leucocytes. There was no evidence that increased heparin clearance contributes to the recovery observed in acute illness. The novel in vivo application of low affinity heparin caused a significant improvement in the recovery and half-life of unfractionated heparin, and the clinical implications of obtaining the benefits of the low molecular weight heparins at a fraction of their cost by using the combination of low affinity and unfractionated heparins are great. Although the results in rabbits may differ quantitatively from those in humans, they are likely to be qualitatively similar, and so it is probable that the variable anticoagulant response to heparin seen in patients is caused by variation in plasma protein binding rather than altered heparin clearance.