BACKGROUND: Several studies have found that musicians perform better than nonmusicians in memory tasks, but this is not always the case, and the strength of this apparent advantage is unknown. Here, we conducted a meta-analysis with the aim of clarifying whether musicians perform better than nonmusicians in memory tasks. METHODS: Education Source; PEP (WEB)-Psychoanalytic Electronic Publishing; Psychology and Behavioral Science (EBSCO); PsycINFO (Ovid); PubMed; ScienceDirect-AllBooks Content (Elsevier API); SCOPUS (Elsevier API); SocINDEX with Full Text (EBSCO) and Google Scholar were searched for eligible studies. The selected studies involved two groups of participants: young adult musicians and nonmusicians. All the studies included memory tasks (loading long-term, short-term or working memory) that contained tonal, verbal or visuospatial stimuli. Three meta-analyses were run separately for long-term memory, short-term memory and working memory. RESULTS: We collected 29 studies, including 53 memory tasks. The results showed that musicians performed better than nonmusicians in terms of long-term memory, g = .29, 95% CI (.08-.51), short-term memory, g = .57, 95% CI (.41-.73), and working memory, g = .56, 95% CI (.33-.80). To further explore the data, we included a moderator (the type of stimulus presented, i.e., tonal, verbal or visuospatial), which was found to influence the effect size for short-term and working memory, but not for long-term memory. In terms of short-term and working memory, the musicians' advantage was large with tonal stimuli, moderate with verbal stimuli, and small or null with visuospatial stimuli. CONCLUSIONS: The three meta-analyses revealed a small effect size for long-term memory, and a medium effect size for short-term and working memory, suggesting that musicians perform better than nonmusicians in memory tasks. Moreover, the effect of the moderator suggested that, the type of stimuli influences this advantage.