Objectives: Develop and implement a music intervention protocol for use with older people with dementia and explore its effect on engagement. Design: A randomised cross-over trial with music intervention and reading control groups. Methods: A music protocol was developed through research team expertise, discussions with facility staff and musicians, practice session feedback and participants' musical preferences. A participant engagement observation checklist was devised to monitor levels of engagement. Results: Participants became more actively (F(2,68) = 5.279, p < .01) and passively (F(2,68) = 4.250, p < .05) engaged in song-singing over time. There was no increase in engagement during the reading control sessions over time. Conclusions: The method by which the music protocol was developed offers a framework and an example for similar, controlled trials. The participant engagement observation checklist also provides an exemplar of how engagement may be captured and illustrates the importance of using one, given participants became more engaged during song-singing.