Speech disorders have become serious health problems in modern society. As a treatment program combining of speech and music, melodic intonation therapy has been used in rehabilitation to stimulate brain functions involved in speech. This is based on two hypotheses: one is that music and language share neurocognitive mechanisms on functional and neural architecture, and the other is that the right cerebral hemisphere of music processing potentially compensates the damaged left hemisphere of language regions. Since Albert, Sparks and Helm (1973) proposed melodic intonation therapy, many adapted versions have appeared based on this version of melodic intonation therapy, and the latter has been named as a traditional version. Since then, both of the traditional and adapted versions were usually used in clinical application. Generally speaking, melodic intonation therapies produce language with an exaggerated prosody, whether the traditional or the adapted versions. For the traditional melodic intonation therapy, the goal of the intervention was to improve patients’ production of propositional language, which required an assemblage of structures according to a set of phonological, morphological, and grammatical rules. The melody always consisted of tones in two different pitches, which were set in a pitch interval of third or fourth. Clinicians ask patients to produce phrases or sentences in a singing-like manner that exaggerates the prosody, meanwhile tapping on each syllable with the left hand. Furthermore, traditional melodic intonation therapy emphasized rigid enforcement procedure and the application scope of patients. Unlike the traditional melodic intonation therapy, the adapted versions aimed to improve the speech production of patients in a manner of melodic intonation by melodic intonation technique. To this aim, the melodies were composed according to speech prosody but not limited in the few tones in two pitches. In addition, the adapted versions were not restrict to a fixed procedure or a certain application scope of patients. The adapted versions could be flexible according to patients’ speech disorders. Although melodic intonation therapy has been applied in different types of speech disorder, such as autism and apraxia, it has been widely applied to the therapy of aphasia, especially for Broca’s aphasia, which was characterized by both linguistic and motor speech impairments. The findings showed that both the traditional and adapted melodic intonation therapies could improve aphasics’ spontaneous speech production, repetition and naming abilities. This suggests the efficacy of melodic intonation therapy in the recovery of aphasia. Brain-imaging evidence showed that the melodic intonation therapies influenced both the structures and functions of human brain. Treatment-associated neural structure changes in patients indicate that the unique engagement of right-hemispheric structures (e.g., the superior temporal lobe, primary sensorimotor, premotor and inferior frontal gyrus regions) and the connections across these brain regions might be responsible for its therapeutic effect. However, with regard of neural functional changes, there existed discrepancy between different results. More specifically, whether recovery from aphasia happen through the recruitment of perilesional brain regions in the left hemisphere, or through the recruitment of homologous language and speech-motor regions in the right hemisphere remained uncertain. This discrepancy may be due to deficiency of experimental control, different task during measurement, inconsistency of patients’ lesion size and different form of melodic intonation therapy. In conclusion, the findings revealed that the melodic intonation therapies improved speech production in aphasia. Future research should pay attention to the underlying mechanisms of the treatment of the melodic intonation therapy. In addition, whether melodic intonation therapy can be effectively applied to mandarin aphasia remains to be further explored.