After a review of the literature on the static fatigue limit, it was concluded that the normal method for its determination, using time-to-rupture data, is inadequate. There are very few experimental demonstrations of the existence of a true fatigue limit, and it is difficult to measure its value with any statistical confidence. An alternative experimental approach is suggested which is based on measuring the change in strength distribution of a representative sample after a high temperature anneal under a fixed tensile stress. This stress weakens weak specimens in the as-received distribution, whereas strong specimens gain in strength. Those specimen whose strength remains constant demonstrate the existence of a fatigue limit and can be used to calculate a numerical value. The results are analysed on the basis of the Charles-Hillig stress corrosion theory, which has been slightly modified to facilitate the examination of the parameters associated with the fatigue limit. This analysis relates the static fatigue limit to the effective surface energy appropriate to the fatigue mechanism, i.e. corrosion caused by water vapour. Some implications of the modified model are discussed.