It has become widely accepted that high strain fatigue is a "delayed static fracture". This idea, which originated in t he work of W.A.Wood (1958), has recently been contradicted by the "random-walk" theory. The latter theory implicitly denied that there is a difference between the phenomena of high and low strain fatigue, as emphasized by Wood. Coffin ' s law of high strain fatigue has been derived from first principles on the basis of the random-walk idea. Gilman (1959) has contributed yet another theory to explain high strain fatigue in terms of the accumulation of vacancies produced by plastic deformation. This dissertation begins with a critical review of the theories, and, since they were based on very limited experimental evidence, data obtained by many different techniques have been used in an attempt to reconcile the contradictions and therefore to advance the understanding of fatigue in general. The internal stress changes caused by high strain fatigue have consequently been investigated, also t he -'damage at the surface where fracture originates, and the fracture surface morphology. The results of this work show that fracture takes place by three processes, namely, crack nucleation, then crack propagation along active slip planes and finally further propagation by a mechanism which causes ripples to form Oh the fracture surface and is not influenced by crystallography. Each process, and its variation through the fatigue strain spectrum, is described in t h is thesis and it is proved that crack propagation involving; ripples is caused by a crack tip rounding and closing mechanism. Because of these findings, the thesis is maintained that the phenomena of high and low strain fatigue are broadly similar and that the existing theories mentioned above are applicable only in very limited fatigue situations. The random-walk derivation of Coffin' s law is shown to be 2 inconsistent with experiment. hypothesis of fatigue failure based on t h e concept that the crack propagates by the production of more surface at the time through localised plastic deformation is proposed in place of the classical theories of fatigue in which it was assumed that the metal at a crB.cl~ tin WB.S workhardened to the exhaustion of its ductility, deacline; to fracture. It is agreed, however, that these theories may be valid in a few special cases. The experimental result s are discussed at the end of each chapter with the intention of introducing the topic of the next. The final chapter contains a general summary and a list of detailed conclusions.