Physical activity is associated with greater independence in old age. While most research has focused on the frequency, intensity and time spent in activity, inconsistent findings have emerged, possibly due to differences across different types of physical activity. Physical activities differ in terms of their non-metabolic, mental, physical and social demands, however, to date, the effects of these demands on functional independence are unexplored. The present thesis aimed to investigate the effect of different types of physical activity on functional independence in old age. A systematic review of randomised controlled trials (RCTs) found a significant, beneficial effect of physical activity on functional performance, with the largest effects found for moderate physical activity levels, and activity types with high levels of mental (e.g. memory, attention), physical (e.g. coordination, balance) and social (e.g. social interaction) demands. These findings generated the hypothesis that physical activities high in mental, physical and social demands are associated with greater functional independence. Thus, a novel typology of physical activities was generated by systematically coding the mental (i.e. attention/concentration, memory, decision-making and strategy), physical (i.e. flexibility, balance, coordination, speeded reactions) and social (i.e. social interaction) demands of 59 physical activities. The typology was then used to recode data from the Understanding Society survey. Findings revealed that as the non-metabolic, physical activity demands increased, functional independence improved. Associations remained significant after controlling for demographics. Key findings included differences across gender, in that the mental and physical demands of activity predicted muscular strength in males, whereas social demands predicted muscular strength in females. In conclusion, physical activities with higher mental, physical and social demands (e.g. dancing) are associated with greater functional independence in old age, compared with simpler types (e.g. walking). Future research is required to test whether these novel findings are replicated elsewhere, ideally using longitudinal or RCT designs.