The beneficial influence of listening to music on anxiety states has often been discussed. However, the empirical evidence and theoretical mechanisms underlying these effects remain controversial. The aim of this study is to conduct a meta-analysis of randomized controlled trials on the effects of music on anxiety in healthy individuals. A comprehensive search in the PsycINFO, Cochrane Library, PubMed and Web of Knowledge databases produced 19 articles complying with the eligibility criteria. The main results of the study reveal an overall decrease in self-reported anxiety (d = −0.30, 95% CI [-0.55,-0.04]); however, the decrease was not significant for psychophysiological signals related to anxiety. Nevertheless, in several cases, listening to music greatly affects blood pressure, cortisol level and heart rate. The great heterogeneity of the studies and the lack of rigorous methodological standards, assessed with CONSORT guidelines, may have biased the results. Thus, listening to music should be cautiously considered as a part of more complex music-based psychological interventions for anxiety regulation. Nonetheless, as discussed in this article, the role of underlying processes (spontaneous memory recollections, mental imagery) must not be neglected. Further research perspectives are discussed.

Music’s power to trigger memories has rarely been tested; in particular, it is not clear what mechanisms govern memory retrieval elicited by musical cues. Previous research has suggested that memory retrieval is underpinned by two mechanisms: (1) distinctiveness—the probability that a cue will retrieve a memory declines with the number of stimuli previously encoded with that same cue and (2) incongruence—a cue encoded with emotionally incongruent targets triggers more memories of the stimuli associated with it than a cue encoded with emotionally congruent stimuli. Our participants experienced an implicit encoding phase where they were presented with auditory-visual pairs of stimuli (pieces of music and images of facial expressions). In the retrieval phase, participants were asked to remember encoded stimuli triggered by music. As expected, musical cues encoded with emotionally incongruent facial expressions triggered more memories than cues encoded with congruent facial expressions. Contrary to our prediction and to previous findings, music from distinctive pairs of stimuli triggered fewer memories than cue pairs displayed multiple times in the encoding phase. Our finding suggests that the manipulation of stimuli at encoding is crucial when using musical cues to trigger memories.