The aim of this research was to investigate the influence of nominal temperature errors on test results and to develop the nominal temperature tolerance for TMF-tests. This concept was confirmed with series of specially developed TMF-tests with defined temperature errors, performed on one test material (NIMONIC 90) in a temperature range of 400850 °C. These tests should show the influence of temperature errors on test results with an emphasis on lifetime in comparison with reference tests. It is shown that with a limited number of specimens and only one test material, it is possible to propose a reasonable general temperature tolerance for TMF-tests, valid for a wide range of materials and test conditions.