Two empirical studies investigate the time course of emotional responses to music. In the first one, musically trained and untrained listeners were required to listen to 27 musical excerpts and to group those that conveyed a similar emotional meaning. In one condition, the excerpts were 25 seconds long on average. In the other condition, excerpts were as short as 1 second. The groupings were then transformed into a matrix of emotional dissimilarity that was analyzed with multidimensional scaling methods (MDS). We compared the outcome of these analyses for the 25-s and 1-s duration conditions. In the second study, we presented musical excerpts of increasing duration, varying from 250 to 20 seconds. Participants were requested to evaluate on a subjective scale how "moving" each excerpt was. On the basis of the responses given for the longer duration, excerpts were then sorted into two groups: highly moving and weakly (or less) moving. The main purpose of the analysis was to identify the point in time where these two categories of excerpts started to be differentiated by participants. Both studies provide consistent findings that less than 1 s of music is enough to instill elaborated emotional responses in listeners.