The three aims of this investigation were (1) to develop a population pharmacokinetic (PK) model for factor VIII (FVIII) in haemophilia A patients, with estimates of inter-occasion and inter-individual variance, (2) to investigate whether appropriate dosing of FVIII for regular prophylaxis can be calculated according to patient characteristics, and (3) to present dosing recommendations for initiating prophylactic treatment. A population PK model was developed using data from four PK studies on patients aged 7-74 years. The model was tested on sparse FVIII data from 42 outpatient visits by haemophilia prophylaxis patients aged 3-66 years. Dose requirements for prophylaxis were calculated both according to the population model and from empirical Bayesian estimates of FVIII PK in the individual patients. The study data were well characterised by a two-compartment PK model. Body weight, age and type of FVIII preparation (plasma-derived or recombinant) were identified as significant covariates. Inter-occasion variance was lower than inter-individual variance for both clearance and volume of the central compartment. The model could reasonably predict FVIII PK in the sparse clinical data. Model-predicted doses (based on age and body weight) to maintain a recommended 0.01 U/mL trough level of FVIII with administration on alternate days started at around 60 U/kg in the small children, decreasing to 10 U/kg or less in middle age. However, "true" dose requirements, as estimated from individual PK parameter data, showed a much greater variation. Appropriate dosing of FVIII for prophylactic treatment cannot be calculated only from body weight and/or age. However, plausible starting doses for most patients would be 1,000 U every other day. FVIII levels should then be checked for dose adjustment.