Research on working memory has suggested domain-specific components for visual, verbal, and spatial information, and more recently for emotion. Affective working memory has been proposed as the set of processes involved in the maintenance of emotions to guide behaviour. The current study examined the reliability of an emotion maintenance/affective working memory task over two experimental sessions separated by one week. Subjective accuracy based on individual ratings was found to correlate over time and was highest for negatively valenced pictures. Results suggest that this paradigm is a reliable measure of emotion maintenance, underscoring the utility of this measure as an assessment tool for normative and clinical populations.