The fatigue properties of three aluminium alloys, DTD683, L65, and AA2024, were determined in reversed bending, with a mean stress of 10 ton/in2, both with and without fretting. Stresses are quoted as σ m ± σ a; i.e. a mean tensile stress of σ m with an alternating stress varying from + σ a (tension) to -σ a (compression). In bending tests only one face of the specimen has a tensile mean stress (σ m), the opposite face having a compressive mean stress (-σ m). However, failure always starts from the face having a tensile mean stress and endurance limits are quoted accordingly. The strength reduction due to fretting was greatest for the high strength Al-Zn-Mg alloy, DTD683, which had an endurance limit of 10 ± 10.3 ton/in2 without fretting but only 10 ± 1.6 ton/in2 under standard fretting conditions, i.e., 1/4 in span fretting pads loaded to 2.5 ton/in2 contact pressure. The strength reduction due to fretting was similar over a range of fretting amplitudes and contact pressures. Fretting will not occur if metallic contact between the mating faces is prevented and tests show that this can be achieved by interposing substances such as graphite, MoS2 or p.t.f.e., in a suitable grease or resinous bonding material, between the mating faces. However, under fatigue loading, greases and resins tended to squeeze out from the contacting faces or break up under the high contact pressures so that they were only effective for a short time. It has been shown that more permanent protection against fretting can be obtained by metal spraying the surface of the specimen, thus preventing direct contact with the fretting pad. Shot peening also produced a significant improvement in fatigue strength under the standard fretting conditions; apparently the sub-surface compressive stresses prevented propagation of the cracks initiated by fretting.