The use of RMS in random load (stress) analysis and generation requires special caution. However, the application of RMS or equivalent stress values to fatigue-life curves obtained under variable amplitude loading leads only in some cases to a good fitting with S—N curves derived from constant amplitude loading. The use of the stress concentration and material related equivalent stress generally gives a better approach than the RMS. As this kind of evaluation may suppress on the one hand maximum values of a distribution as well as exceedances of the endurance limit, and on the other hand may invert the damaging influence depending on the type of distribution, it bears several disadvantages in the presentation of variable amplitude fatigue data. The presentation in form of the respective maximum value of the cumulative frequency distribution is superior because the maximum values can be compared with static failure data, allowable exceedances of the endurance limit, can be evaluated and the damaging influence of differemt distributions can be recognized.