Studies of Factor VIII pharmacokinetics in haemophiliacs can be classified into 2 groups depending on whether single-dose or multiple-dose Factor VIII curves are used. This review analyses information published so far in both these areas, with particular emphasis on the choice of appropriate models for pharmacokinetic analysis. Single-dose studies of Factor VIII kinetics have previously used a wide variety of methods for pharmacokinetic analysis (empirical methods of Factor VIII level prediction, graphical techniques for semilog analysis, 1-compartment and 2-compartment models). However, Factor VIII poses unique problems to the pharmacokineticist because decay curves can be either monophasic (monoexponential) or biphasic (biexponential) for unknown reasons, and because Factor VIII concentrations are generally subject to significant assay error. Problems of compartmental analysis that occurred in previous studies are highlighted, and a model-independent non-compartmental approach for analysing Factor VIII curves is proposed. To date, fewer data have been published on multiple-dose kinetics of Factor VIII. From a clinical point of view, repeated-dose regimens are most commonly required in patients undergoing surgery and in patients with severe bleeding. A fairly well defined 'therapeutic window' of optimal Factor VIII plasma concentrations has been identified, particularly in surgical patients. This fact has spurred research aimed at applying to haemophilia patients the pharmacokinetic dosing methods commonly used for therapeutic monitoring of drugs (e.g. Bayesian method for dosage individualization). A few papers have already been published in this field, and this review summarises problems encountered by previous investigators, and evaluates comparatively the pharmacokinetic methods used.