Twelve existing lifetime models for optical fibers, based on a power law to describe the stress-enhanced crack growth, are studied and compared. The number of models is reduced to only one basic model. This model takes into account the effects of a proof test. An alternative model, to be used when proof testing is performed on line, is given as a worst case limit. A choice of three testing methods to obtain information about the weak-flaw distribution is given: dynamic fatigue and variable screen testing of long lengths or the use of a failure number during the proof test. We show that the part of the crack growth that is controlled by transport of the reaction agent instead of by reaction speed (described by the power law) does not influence the model. It is also shown that weakening of the fiber during unloading has a minor effect on the presented model. The lifetime model cannot be used for fibers in water without further study.