The aim of this master thesis is to set-up an automatic Rainflow Counting of load-time functions according to the international standard ASTM E 1049 85 within the software package IBF-Fatigue by Ingenieurbuero Fiedler GmbH based on FE stress results. The various methods of the one- and two-parameter counting procedures are examined in the state-of-the-art with regard to their advantages and disadvantages for the counting of the stress-time functions on the basis of stress range and mean values as well as upper and lower loads. For the preparation of the selected method Rainflow Counting within a Mathcad file, a calculation process was created for both a residual of 0 and 0.5 as well as for a shifted line Max_Max. The validation of the function and results by the software tool was conducted by a comparison with manually evaluated results. An analysis of the influence of the residuals show that the main difference between the results without and with considering the residuals depends on both the position of the largest peaks and the number and difference in values between the largest peak and other values. For any load-time function class 64 is feasible with high reliability and class 32 may be selected with reservation. An optional function with selectable tolerance was added to the software to reduce unnecessary small vibration range, which can speed up the processing process, albeit with a small error. Finally, the results according to the rainflow-exact load spectra are compared with the results of the rainflow-classified load spectra (class 64) by both a residual of 0.5 and the displacement Max_Max in the software tool. This was performed on the basis of test calculations of the FKM guideline for load spectra on welded steel construction components of an exemplary heavy machine with different FAT classes. If calculating according to the FKM guideline, a deviation of up to 2% In the same FAT-class occurs, mainly due to the classification of the load spectra between the exact and the classified load spectra with the residual of 0.5. The shifted curve- Max_Max is up to 5% more reliable compared to the original curve.