There are many challenges facing epidemiologists wishing to investigate relationships between physical activity, obesity, diabetes, dementia and depression, all of which are complex fields in their own right. There is a large literature investigating the relationship between diabetes and dementia but less, as yet, on the other exposures and outcomes. In this literature there is a diversity of definitions making rigorous systematic review problematic. There is a need to define hypotheses in this area very clearly and to identify studies that have addressed the specific question. Such exercises have not been carried out to date but would enlighten the research area and point more clearly to questions which remain to be answered. Our own research group has examined the specific question of risk of development of dementia in relation to levels of HbA(1)c, as a marker of glycaemic control and showed that although not related to dementia, it is related to incidence of severe cognitive impairment.