The object of this research was to establish the mechanism of corrosion fatigue by the aid of chemical and electrochemical measurements, with special reference to the possibility of preventing corrosion fatigue by means of cathodic currents. Two-stage tests on steel specimens subjected to alternating stress, with a chloride solution applied during the first stage only, have indicated that as the first stage (corrosion period) is increased, the total life, at first immeasurably long, becomes extremely short and then increases again. The unexpected increase in total life produced by an extension of the corrosion period may be explained by the fact that isolated cracks produce more stress intensification than a number of neighbouring cracks. Another series of experiments has shown that the rate of passage of iron into the combined state greatly increases with the applied stress. The application of a cathodic current diminishes the rate of production of iron compounds and the number of cracks; weak cathodic currents actually shorten the life, but still stronger ones increase itagain-which can be explained in two ways: (1) for a given depth few cracks cause more weakening than many, (2) for a given amount of corrosion, an increased number of cracks means smaller depth of cracking and hence less damage. If the current reaches a certain value, corrosion becomes undetectable, and the life becomes extremely long in neutral potassium chloride (but not in acid). The value of the current needed for this complete protection increases with the stress range, but the value of the potential corresponding to the protective current moves steadily lower, i.e. in the direction of zinc, with applied stress. Applying the graphical methods of representing corrosion phenomena to these results, we are led to the view that at least three different factors operate in causing alternating stress to enhance the rate of corrosion and the rate of mechanical damage. These are (A) diminution of cathodic polarization, (B) diminution of anodic polarization, and (C) diminution of the resistance of the path joining anodes and cathodes. It is possible that there may also be (D) a bodily shift of the anodic polarization curve in the base metal direction. Studies of the shift of potential with time in presence of different types of stress indicate that stressing within the elastic range only affects the potential by altering the state of repair of the film covering the surface. It is likely that stresses within the plastic range depress the potential of the metal itself-irrespective of any damage to a film-but further work will be needed definitely to establish this point.