Pre-clinical studies in mice and haemophilic dogs have shown that introduction of an adeno-associated viral (AAV) vector encoding blood coagulation factor IX (F. IX) into skeletal muscle results in sustained expression of F. IX at levels sufficient to correct the haemophilic phenotype 1, 2. On the basis of these data and additional pre-clinical studies demonstrating an absence of vector-related toxicity, we initiated a clinical study of intramuscular injection of an AAV vector expressing human F. IX in adults with severe haemophilia B. The study has a dose-escalation design, and all patients have now been enrolled in the initial dose cohort (2× 10 11 vg/kg). Assessment in the first three patients of safety and gene transfer and expression show no evidence of germline transmission of vector sequences or formation of inhibitory antibodies against F. IX. We found that the vector sequences are present in muscle by PCR and Southern-blot analyses of muscle biopsies and we demonstrated expression of F. IX by immunohistochemistry. We observed modest changes in clinical endpoints including circulating levels of F. IX and frequency of F. IX protein infusion. The evidence of gene expression at low doses of vector suggests that dose calculations based on animal data may have overestimated the amount of vector required to achieve therapeutic levels in humans, and that the approach offers the possibility of converting severe haemophilia B to a milder form of the disease.