The hot torsion test enables the analysis and simulation of hot forming up to large strains. The gradients of temperature, strain and strain rate occurring in the specimen result in complex material behavior. The aim of this work is to gain a better understanding of the test conditions, and to implement an evaluation for the test. For this purpose, torsion tests were carried out on an austenitic steel (A220) in a Gleeble 3800. As part of the evaluation, a finite element model was implemented in ABAQUS®. The parameterization of the Hensel-Spittel constitutive model was done by an inverse analysis of the experimental torsion data. Starting from an initial guess, the parameter set with the smallest deviation between numerical and experimental data is searched for. This optimization problem was solved using the Nelder-Mead algorithm. In the experimental results it has been shown that certain experimental conditions promote a local instability. This could lead to a significant temperature rise, which results in flow localization. An evaluation using the classical analytical method can then no longer be carried out reliably. The evaluation carried out via an inverse analysis, has shown a good agreement with the experimental results. The comparison of the determined constitutive parameters were in good agreement with literature data from compression tests.