The goal of the presented research is to experimentally investigate the influence of reversed-cyclic pure shear loading on the behaviour of reinforced-concrete elements, and to formulate simple analytical models which capture the defining features of non-monotonic shear behaviour. This paper details the reversed-cyclic pure shear testing of three specimens containing varying amounts of transverse reinforcement. The results demonstrate that using monotonic predictions for reversed-cyclic shear strength can be significantly unconservative, with strengths as much as 25% lower than monotonic predictions. A fixed-crack model which accounts for crack-closing behaviour is introduced in order to model the load path-dependent behaviour of these experiments.