Verification of vehicle components with respect to fatigue is discussed. The two most common methods, i.e. the safety factor approach and the load-strength method are studied. A set of problems with the methods is identified. An alternative using the advantages of the two techniques is developed. This combined approach uses a signal-to-noise ratio, which can be related to classical safety factors. It is then possible to evaluate the performance based on probability theory as well as updating in a rational way when new observations are obtained. Assumptions of the distribution tails are not necessary, but can be replaced by a rough judgement based on experience and field failure statistics. A set of measurements on steering knuckles is used to exemplify the method.
