There are a number of limitations to this research. From this study, it is not possible to establish whether it was the physical exercise programme itself or the social aspects of the intervention that led to improved function. This is important as there is evidence to suggest that social interventions may be effective in reducing some of the symptoms of dementia (36). Further studies utilizing control interventions with a comparable social component would help to clarify this issue. A second limitation is that the study was not double blinded. While the outcome assessor was blinded, participants and their carers knew if they were in the intervention (exercise) group or not, and this may have led to an expectation of improvement amongst those who exercised and an expectation of decline in those who did not. Again, a control intervention would assist to overcome this in future studies. Finally, the intervention and follow-up periods were short (4 months). It would be interesting to conduct an extended trial with longer follow-up to see whether the short-term benefits were maintained.