This review considers how research into the impact of music upon people with dementia has been conducted, with a focus on the means used to measure outcomes Although there is a rapidly-growing amount of research in this relatively new area, it has evolved fairly separately within different fields, such as psychology, music therapy and nursing. Using principles from the meta-narrative review method, this paper systematically surveys qualitative and quantitative measures which have been used in studies relating to music and dementia from 1980 to the present day. Data extracted included the outcome measures used, the type of music intervention, the setting of the studies and methodology and study design. Results showed that quantitative studies greatly outnumber qualitative studies, and that most research takes place in long term care settings. Instruments which measure neuro-psychiatric symptoms of dementia, especially agitation, were the most frequently used. Other domains that were targeted by the studies included quality of life, physiological change, music-related outcomes and cognitive skills; the inter-relations between these domains are represented as a diagram. Researchers' rationales for choosing to target these outcomes are discussed. The conclusion suggests that, although reduction of neuropsychiatric symptoms is a valuable outcome, it is important that researchers are open to the possibility of other outcomes. We suggest that measuring increase in positive responses might be an alternative to measuring reduction of negative symptoms, and highlight the importance of listening to people with dementia and carers when designing studies.