OBJECTIVE: To assess whether identifying (or ignoring) learned alarm sounds interferes with performance on a task involving working memory. BACKGROUND: A number of researchers have suggested that auditory alarms could interfere with working memory in complex task environments, and this could serve as a caution against their use. Changing auditory information has been shown to interfere with serial recall, even when the auditory information is to be ignored. However, previous researchers have not examined well-learned patterns, such as familiar alarms. METHOD: One group of participants learned a set of alarms (either a melody, a rhythmic pulse, or a spoken nonword phrase) and subsequently undertook a digits-forward task in three conditions (no alarms, identify the alarm, or ignore the alarm). A comparison group undertook the baseline and ignore conditions but had no prior exposure to the alarms. RESULTS: All alarms interfered with serial recall when participants were asked to identify them; however, only the nonword phrase interfered with recall when ignored. Moreover, there was no difference between trained and untrained participants in terms of recall performance when ignoring the alarms, suggesting that previous training does not make alarms less ignorable. CONCLUSION: Identifying any alarm sound may interfere with immediate working memory; however, spoken alarms may interfere even when ignored. APPLICATION: It is worth considering the importance of alarms in environments requiring high working memory performance and in particular avoiding spoken alarms in such environments.