Controversies and debates remain regarding the best management of severe acute‐onset autoimmune hepatitis (SA‐AIH) due to the lack of useful outcome or complication prediction systems. We conducted this clinical practice‐based observational study to clarify whether Chronic Liver Failure Consortium Organ Failure scores (CLIF‐C OFs) and the computed tomography–derived liver volume to standard liver volume (CTLV/SLV) ratio at admission to a tertiary transplant center can predict outcomes and complications due to infection. Thirty‐four consecutive corticosteroid‐treated patients with SA‐AIH from 2007 to 2018 were included. Severe hepatitis was defined as an international normalized ratio (of prothrombin time) over 1.3 any time before admission. Of the 34 corticosteroid‐treated patients with SA‐AIH inclusive of 25 (73.5%) acute liver failure cases, transplant‐free survival was observed in 24 patients (70.6%). Any infection was noticed in 10 patients (29.4%). CLIF‐C OFs, at the cutoff of 9, significantly predicted survival (P = 0.0002, log‐rank test), outperformed the Model for End‐stage Liver Disease system in predicting outcome (P = 0.0325), and significantly discriminated between liver transplant and death in a competing risk analysis. SA‐AIH was characterized as having decreased CTLV/SLV, which was also predictive of survival (P < 0.0001). Interestingly, CLIF‐C OFs, especially the subscores for respiratory dysfunction, also predicted infection (P = 0.007). Conclusion: In corticosteroid‐treated patients with SA‐AIH, CLIF‐C OFs and CTLV/SLV ratios predicted both survival outcome and complications due to infection. Further investigation is warranted to determine whether making decisions based on CLIF‐C OFs or CTLV/SLV ratios is useful.