Corticosteroids alone or in conjunction with azathioprine (AZA) is the standard treatment in autoimmune hepatitis (AiH). Individual variations in thiopurine (TP) metabolism may affect both drug efficacy and toxicity. Our aim was to investigate the utility of thiopurine methyltransferase (TPMT) as well as thioguanine nucleotide (TGN) and methylthioinosine monophosphate (meTIMP) metabolite measurements with regard to clinical outcome.Two hundred thirty-eight patients with AiH were included in this cross-sectional study. TPMT status was assessed in all patients, while TGN and meTIMP were measured in patients with ongoing TP medication. Clinical outcome was evaluated by liver tests and the ability to withdraw steroids.TPMT genotyping (n=229) revealed 207 (90.4%) wild-type and 22 heterozygous patients. One hundred forty-three patients had ongoing TP therapy with AZA (n=134) or mercaptopurine (MP; n=9); response was judged as complete response (CR) in 113 patients and partial response (PR) in 30 patients. Both TP dose (1.64 vs 1.19 mg/kg; p=0.012) and TPMT activity (14.3 vs 13.5; p=0.05) were higher in PR, resulting in similar TGN levels (PR: 121 pmol/8 x 10(8) red blood cells [RBC]; CR: 113 pmol/8 x 10(8) RBC; p=0.33) but higher meTIMP levels in PR (1350 vs 400 pmol/8 x 10(8) RBC; p=0.004). Patients able to withdraw steroids or who were using 5 mg prednisolone daily were treated with lower TP doses than patients on higher steroid doses (1.15 vs 1.18 vs 1.82 mg/kg; p<0.001).TP metabolite measurements are of clinical value in AiH patients who do not respond to standard TP treatment and for the identification of a shifted metabolism, which may demand an alternative treatment strategy.