A bearing service life prediction methodology and tutorial indexed to eight probable causes for bearing failure and removal are presented - including fatigue. Bearing life is probabilistic and not deterministic. Bearing manufacturers' catalogue (L10) bearing life is based on rolling-element fatigue failure, at which time 90% of a population of bearings can be reasonably expected to survive, and 10% to fail by fatigue. However, approximately 95% of all bearings are removed for cause before reaching their L10 life. A bearing failure can be defined as when the bearing is no longer fit for its intended purpose. For a single bearing, you can only predict the probability of a failure occurring at a designated time - but not the actual time to failure.

When a power transmission component fails, it can adversely affect the performance of the assembly, often making the machine inoperable. Such failures can not only harm the reputation of the manufacturer, but can lead to litigation, recalls and delays in delivery due to quality concerns. Some failures can even result in bodily injury or death. Understanding why a part failed is critical to preventing similar failures from reoccurring. In the study of a failed part, the analyst must consider a broad range of possibilities for the failure. Although some failures can be attributed to a single primary cause, it is common for multiple secondary factors to contribute. The failure analyst must evaluate all of the evidence available to prepare a hypothesis about the causes of failure.

Dovetails, gears and splines have been widely used in aero engines where fretting is an important failure mode due to loading variation and vibration during extended service. Failure caused by fretting fatigue becomes a prominent issue when service time continues beyond 4,000 hours. In some cases, microslip at the edge of a contact zone can reduce the life by as much as 40–60 percent.