Previous studies suggest divergent effects of emotions on memory functions: impairing associative memory while enhancing item memory (e.g. Bisby & Burgess, 2014; Bisby et al., 2018). Working memory is considered a potential mechanism underlying the impairing effect of emotion. On one hand, arousal is claimed to hinder working memory, complicating the association of emotional and neutral stimuli. On the other hand, it is suggested that unique features of emotional stimuli may elicit diverse brain activations, potentially reducing working memory performance. The question is whether emotional disruption of associative memory arises from its impact on working memory or from shared activation with other brain regions, like visual areas, impeding working memory activation during emotional stimulus association. This study addresses this question by manipulating working memory performance under high and low cognitive load conditions. Participants encoded pairs of neutral (neutral scene-neutral object) and emotional (emotional scene-neutral object) stimuli, and then solved either easy or difficult math problems after each trial. Subsequently, they underwent an associative memory test immediately or 24 hours later. If the first suggestion is valid, emotional stimuli should show no retention interval effect. If the second suggestion is valid, increased working memory activation would enhance associative memory. The results demonstrated that high cognitive load improved immediate associative memory performance for neutral pairs and delayed performance for emotional pairs. In conclusion, boosting working memory enhanced associative memory. This effect persisted over time for emotional pairs but diminished for neutral pairs. Results supported the second suggestion over the first.