Torsion tests allow to study the rheological properties of various materials, including properties in hot state, as well as they allow to conduct physical simulations of real material forming processes, including processes of intensive and alternating deformation. However, it is often a question what size the specimen should have to perform torsion test accurately. The article aims to study the influence of cylindrical specimen size on the distribution of deformation during torsion test. For this purpose, computer simulation of the torsion testing process was performed. The influence of the relative length of the specimen gauge on the deviation of the actual values of the effective strain from the calculated values was quantified. It is shown that the error in the calculation of the effective strain, based on the values of the twist angle and the specimen gauge size, is of 4%. In order to verify the results obtained by computer simulation, the physical simulation was performed. It is shown that the distribution of deformation along the length of the specimen gauge is also significantly influenced by the accuracy of specimen manufacture, as well as the specimen material. In some cases, the error in the calculation of the effective strain, based on the values of the twist angle and the specimen gauge size, can reach more than 80%.