Music is one of the most sublime stimulus that human beings can experience. Despite being just an acoustic wave that exert little physical action on a subject, it triggers profound changes in emotions and physiological states. This study explores the possibility of detecting subtle changes in cerebral blood flow velocity in response to emotional reactions produced by different musical stimuli using multiscale entropy analysis. This method has shown to be able to reveal the complexity of internal dynamics of different physiological systems, which cannot be appreciated with classic approaches based on entropy on a single scale. Cerebral blood flow signals were successfully recorded for 16 subjects while performing five different music tasks. The entropy of each signal was estimated using multiscale sample entropy. Significant differences in entropy was found between two tasks, which suggests that intense cognitive activities having emotional content yield a decrease in the entropy of cerebral haemodynamics.