Since the 1960's polysomnographic sleep research has demonstrated that depressive episodes are associated with REM sleep alterations. Some of these alterations, such as increased REM sleep density, have also been observed in first-degree relatives of patients and remitted patients, suggesting that they may be vulnerability markers of major depressive disorder (MDD), rather than mere epiphenomena of the disorder. Neuroimaging studies have revealed that depression is also associated with heightened amygdala reactivity to negative emotional stimuli, which may also be a vulnerability marker for MDD. Several models have been developed to explain the respective roles of REM sleep alterations and negatively-biased amygdala activity in the pathology of MDD, however the possible interaction between these two potential risk-factors remains uncharted. This paper reviews the roles of the amygdala and REM sleep in the encoding and consolidation of negative emotional memories, respectively. We present our ‘affect tagging and consolidation’ (ATaC) model, which argues that increased REM sleep density and negatively-biased amygdala activity are two separate, genetically influenced risk-factors for depression which interact to promote the development of negative memory bias – a well-known cognitive vulnerability marker for depression. Predictions of the ATaC model may motivate research aimed at improving our understanding of sleep dependent memory consolidation in depression aetiology.