Background <p>Cyclic loading is a leading cause of material failure and accurate testing methods are required to predict the life of components.</p> Objective <p>This work investigated the challenges of performing reversed three-point bending experiments for characterising the fatigue life of polymers and composites, and sought to develop guidelines for improving the testing method.</p> Methods <p>Testing was performed with polypropylene (PP) and glass-fibre reinforced PP on a test fixture that featured double-sided supports. Monotonic and fully reversed (i.e. stress ratio <i>R</i> = - 1) cyclic bending experiments were conducted to characterise the effects that the test conditions, including the type of supports and level of contact between the specimens and anvils, have on the measurements.</p> Results <p>Stresses generated during reversed bending are highly dependent on the type of fixture configuration used. Most configurations that were tested led to mechanical interlocking that altered the bending moments and generated membrane stresses. In this work, the purest form of reversed bending was achieved using double-sided pivoting supports, and a 0.35 mm gap between specimens and anvils to allow sliding in the supports.</p> Conclusions <p>This configuration provided the most similar stress state to that obtained from the standardised simply-supported beam experiments that are used for characterising the quasi-static flexural properties of materials, and was ultimately the configuration that provided the most reliable fatigue data.</p>