OBJECTIVES The assumption that working memory (WM) is embedded within long-term memory suggests that the effectiveness of switching information between activated states in WM (i.e., attentional refreshing) may depend on whether that information is semantically relevant. Given that older adults often have greater general knowledge than younger adults, age-related deficits in episodic memory (EM) could be ameliorated by studying information that has existing semantic representations compared with unknown information. METHOD Younger and older adults completed a modified operation span task that varied the number of refreshing opportunities. The memoranda used were equally known to younger and older adults (neutral words; e.g., father), better known to older adults than younger adults (dated words; e.g., mirth), or unknown to both groups (unknown words; e.g., cobot). RESULTS Results for immediate and delayed recall indicated an age-related improvement for dated memoranda and no age difference for unknown memoranda. Furthermore, refreshing opportunities predicted delayed recall of neutral memoranda more strongly for younger adults than older adults, whereas older adults' recall advantage for dated memoranda was explained by their prior knowledge and not refreshing opportunities. DISCUSSION The results suggest that older adults' EM deficits could potentially be ameliorated by incorporating their superior knowledge to supplement relatively ineffective attentional refreshing in WM.